Want to feel what the night was like? From the talks to the networking and everything in between, here's the aftermovie from Qwen Meetup Seoul #4 🎥.

On August 19 we ran Qwen Meetup Seoul #4 together with Qwen at dcamp Seolleung. Around 130 people came out for the evening.

Both talks were about problems that become a lot more obvious once you start using AI seriously: first, how expensive coding agents can get when you're running frontier models for everything, and second, what happens when the AI agent you built stops being a demo on your laptop and has to survive in production.

Different layers of the stack, but actually a pretty similar question underneath both:

How do you make AI systems efficient enough to keep using them at scale?

You probably don't need your most expensive model for everything

Edwin Tack, Solutions Architect at Alibaba Cloud Korea, opened with a question for the room: what are people actually using for coding every day?

A lot of hands went up for Claude Code. Some for Codex. Others were using models like DeepSeek.

His starting point was simple. Model performance matters, obviously, but once every engineer on a team starts running agents all day, the cost starts mattering too.

And maybe the answer isn't to choose one model.

Edwin split an AI coding workflow into two broad stages:

Planning and design, where you want the strongest reasoning you can get.

And execution, where the agent is spending much more of its time actually writing, modifying and iterating on code.

The proposal was to keep a frontier model for the planning stage, but use something cheaper such as Qwen3.8 Max for execution.

Instead of asking which model is best, ask which model is best for this particular step.

According to the estimates Edwin presented, a hybrid setup could cut model costs by roughly 30 percent, while moving both planning and execution to Qwen could push the reduction closer to 50 percent.

[Photo: slide showing the hybrid model / cost comparison]

I liked the analogy he used later in the talk.

Infrastructure teams rarely assume one cloud provider must run absolutely everything. Having AWS, Azure, GCP, Alibaba Cloud or another provider competing for different workloads gives companies options.

His argument was that AI models are heading in the same direction.

We shouldn't necessarily think of Claude, GPT, Qwen, GLM, Kimi or DeepSeek as one permanent choice. The application can route work between them depending on quality, latency and cost.

That becomes even more interesting as the performance gap between models gets smaller.

The model gap is shrinking

The Q&A went quite deep into this.

One question was basically: cost is nice, but time is money too. If Qwen takes longer to produce the same result, are you actually saving anything?

Edwin's view was that the gap between frontier models and the models following them is getting increasingly small, especially for many coding workloads.

New models keep arriving, benchmarks keep moving, and something that looks comfortably ahead today might have several close competitors a few months later.

His prediction was that cost will therefore become a bigger part of the decision.

Another attendee asked what happens to AI pricing over the next few years. Edwin expects the most powerful frontier models to remain expensive, while cheaper competitors put pressure on how far those prices can actually rise.

That leads back to the hybrid approach: spend the expensive tokens where they make the biggest difference, and don't spend them where they don't.

There were also questions about running smaller Qwen models locally, passing context between models, and how to plug this kind of setup into existing coding tools.

The nice part was that the discussion wasn't really "Qwen versus everything else."

It was much more practical:

If I already use the models I like, where else in my workflow could another model make sense?

Link: How to optimize Vibe coding costs with Qwen3.8 Max — full talk

Your agent isn't a chatbot

Then PD, Senior Solution Architect at TiDB, took the stage and started somewhere I definitely wasn't expecting: with a camera he lost in the sea in Thailand 17 years ago.

He had been climbing back onto a small boat after snorkeling when the camera hanging around his neck got caught and dropped into the water.

The camera itself wasn't what bothered him.

The photos were.

A month later, somehow, a diver found it around ten meters underwater. Someone posted the recovered photos online, one of PD's friends recognized him, and he eventually got the camera and his memories back.

He still has those photos today.

That was his setup for the rest of the talk.

The valuable part of an AI agent often isn't the process itself.

It's the state the process accumulated while working.

And unlike a simple chatbot, an agent might have a whole sequence of work to complete:

fetch the order, check the policy, calculate something, call another API, issue the refund, update the result.

Now imagine the process dies halfway through.

If all of that state lived only inside the running process, the new process doesn't know what already happened.

It starts again.

When restarting means paying twice

The harmless version of that problem is wasted tokens.

An agent gets through six steps, the pod dies, it restarts from step one, and you pay again for work it already completed.

PD demonstrated exactly that.

In his example, restarting the workflow from scratch used roughly 2,300 tokens in total. Once the agent wrote a checkpoint into TiDB after each completed step, it could recover from the last successful checkpoint instead.

The same interrupted workflow finished at roughly 1,300 tokens.

But tokens aren't the scary part.

Imagine the action halfway through the workflow was a refund.

The agent checks that the customer qualifies, calculates the amount, sends the refund, and then crashes before remembering that the refund was sent.

It starts over.

Congratulations to the customer. They may now have two refunds.

PD's point was that we've spent decades learning how to make financial and transactional systems reliable, and then started building agents that keep important state inside a Python process.

Putting an LLM in the loop doesn't remove the old distributed systems problems.

It adds new ways to trigger them.

The prototype worked. Then users arrived.

This was probably the most useful distinction in the talk.

Building an agent prototype today is surprisingly easy.

You have your model, maybe PostgreSQL or MySQL for application data, a vector database for embeddings, Redis for some state, and APIs connecting everything.

It works.

You demo it after a weekend and look like a genius.

Then you get real traffic.

Processes run out of memory. Pods restart. Deployments replace instances while agents are halfway through a task. Hundreds or thousands of workflows start writing state concurrently.

Meanwhile your transactional database has one version of the data, your vector database has another, and the embedding pipeline hasn't caught up with yesterday's policy change yet.

Those are problems you don't see when ten developers are testing the prototype.

You see them when customers arrive.

PD described TiDB's role as a persistent layer underneath the agents: keep checkpoints in a normal SQL database, recover workflows after failures, and handle the concurrency once the application grows past the point where a single database node is comfortable.

The larger architecture he showed combines transactional data, vector search, full-text search and analytical workloads rather than maintaining a separate database and synchronization pipeline for every one of them.

Yes, you can do checkpointing with PostgreSQL

Someone eventually asked the question PD knew was coming:

Couldn't you just do this with PostgreSQL?

His answer was yes.

The checkpoint itself isn't magic. You can write a row containing the task ID, completed step and relevant state into almost any database.

His argument for TiDB starts when the workload becomes large.

For a small application, he was very clear that PostgreSQL can actually be faster because the architecture is simpler.

But once concurrency keeps increasing, one primary writer eventually becomes a bottleneck. TiDB distributes writes across nodes and can add capacity horizontally without requiring the application to be redesigned around a new database architecture.

That distinction came up several times during the Q&A:

TiDB isn't supposed to win the tiny benchmark. It's supposed to keep behaving predictably after the tiny benchmark stops resembling your production system.

There were good follow-up questions on Raft consensus, write latency, strong versus eventual consistency, storage tiering, embedding models, and how TiDB can serve both transactional and analytical workloads.

Link: When AI apps become real: The hidden infrastructure behind agents in production — full talk

Two different talks, the same constraint

Looking back, the talks connected better than I expected.

Edwin's talk was about not wasting your most expensive model on work that a cheaper model can do.

PD's talk was about not wasting tokens doing work an agent already completed.

One optimizes which model performs the work.

The other makes sure you don't perform the same work twice.

As agents become a normal part of software products, those details matter more.

A demo can ignore a few extra dollars in tokens. It can ignore the occasional restart. It can use one model for everything and keep state wherever is easiest.

A production system with thousands of users can't.

The room

Around 130 people joined us this time.

We started with food and check-in, moved into the two talks and Q&As, and then kept the room open for networking afterward.

One thing I've been enjoying about these AI meetups is how technical the conversations continue to get after the official program ends. You hear people comparing models, architectures, tools and things they're actually building rather than just talking about AI at a high level.

That's what I want these events to be.

Huge thanks

This event wouldn't have happened without:

• Edwin Tack for breaking down where a multi-model coding workflow can actually save money
• PD for turning a lost camera into a surprisingly good explanation of persistent agent state
• Soo & Febria for continuing to build the Qwen community in Seoul with us
• dcamp for hosting us at Seolleung
• Cry Cheeseburger for the burgers
• The volunteers who kept the night running: Bertijn, Brian, Nathaniel, Oscar, and Suhyun
• And all 130 of you who came, asked unusually technical questions, and stayed around afterward to keep the conversations going

Watch the talks

Both full talks are online:

• How to optimize Vibe coding costs with Qwen3.8 Max — Edwin Tack
• When AI apps become real: The hidden infrastructure behind agents in production — PD

You can also find every published Dev Korea talk on our talks page.

See you at the next one.

Ready for your next move?
Visit Dev Korea to explore the latest job openings at dev-korea.com/jobs, or if you're hiring, post a job at dev-korea.com/post-a-job and connect with our growing international tech community.

Moving house is obviously not something unique to Korea. But having now done it here, there were quite a few parts of the process that felt distinctly Korean, from having a moving company pack and unpack almost everything for us, to the huge ladder truck outside our old apartment, putting down protective covering in the new apartment, my mother-in-law putting red beans in the corners of the apartment, and a final trip to the local 주민센터 (jumin center) to update my address.

So rather than trying to write a definitive guide to moving in Korea, I thought I'd document what our moving day actually looked like.

Hopefully it gives you an idea of what to expect if you're moving apartments here for the first time.

We chose 포장이사, and I'm very glad we did

For the move, we chose what is called 포장이사 (pojang-isa) in Korean.

The easiest way to describe it is a full-service move. The moving company doesn't just transport your furniture from point A to point B. They pack your belongings at the old apartment, move everything, and then unpack and put things back at the new place.

And when I say they pack everything, I really mean almost everything.

Clothes, kitchen items, books, furniture. The team went through the apartment and packed everything into their own boxes and containers.

For our move, the total cost ended up being 900,000 KRW.

We paid a 50,000 KRW deposit when booking, followed by 800,000 KRW for the move itself. We then paid an additional 50,000 KRW because the movers had to wait around 1.5 hours before they could start unloading at the new apartment.

That extra waiting fee wasn't mentioned in our contract. In our case, the delay happened because the concierge at the new building was away for lunch and we couldn't start moving things in until they returned.

It's probably something worth keeping in mind when planning a move in Korea. Even if the moving quote itself is fixed, delays at either building can potentially result in an additional charge.

In our case, we were moving from approximately 39 m² to 43 m².

I have quite a bit of tech equipment, and one thing I specifically wanted them to be careful with was my motorized standing desk. It's large, made up of several parts, and fairly expensive, so I asked them to take extra care with it.

They did. The desk and the rest of our furniture were packed very carefully, and anything that needed extra protection was properly wrapped before being moved.

The part that impressed me most came at the other end.

They didn't simply unload dozens of boxes into our new apartment and leave us with several days of work.

They actually unpacked things.

Clothes went back into closets and drawers. Furniture was placed where we wanted it. Items were brought back into the appropriate rooms, and the movers paid attention to how things had been organized in the previous apartment.

Of course, that doesn't mean the apartment magically looked finished at the end of the day.

I still had this waiting for me:

Apparently even 포장이사 has its limits.

But compared with packing, carrying and unpacking an entire apartment ourselves, the service made the whole move dramatically easier.

What the day actually looked like

The movers arrived at our old apartment at around 8:00am.

By roughly 11:00am, they had finished packing everything and clearing out the apartment.

Because we were renting under a jeonse contract, the next hour wasn't really about moving. From around 11:00am to noon, we were at the real estate agency handling the paperwork for the end of the lease and getting our jeonse deposit back.

After that, we took a taxi to the new apartment, which was about a 30-minute drive.

We arrived around 12:30pm, but had to wait until about 1:00pm because the concierge at the new building was still on lunch break.

Once we could start, the movers brought everything up to the 19th floor, unpacked, and started putting things back in place.

They were completely finished by around 3:00pm.

So the actual moving work took around five hours in total, split between the morning packing at the old apartment and the afternoon unloading and unpacking at the new one.

Considering that they packed almost everything for us, moved the furniture, used the ladder truck at the old apartment, brought everything up to the 19th floor at the new place, and then unpacked and reorganized a lot of it, that felt incredibly efficient.

My mother-in-law arrived around 4:00pm, which is when the red beans started appearing in the corners.

Yes, there was a giant ladder outside our apartment

One of the most visually interesting parts of moving in Korea is the 사다리차 (sadaricha).

It's essentially a truck with a long exterior lift that can reach the windows or balconies of apartments several floors above the ground.

Our previous apartment was on the 4th floor, so the movers could use the ladder truck there. Instead of carrying everything through the building, they loaded boxes and furniture onto the platform outside the apartment and sent them directly down to the truck.

Our new apartment is on the 19th floor, so the ladder truck wasn't an option when we arrived. Everything had to be brought up through the building instead.

That made the contrast pretty interesting. At the old place, our belongings were literally disappearing out of the window. At the new place, the team had to handle the unloading much more traditionally through the building.

If you've spent much time walking around residential neighborhoods in Korea, you've probably seen these trucks before. They're a very familiar sight on Korean moving days, especially around apartment buildings.

Arriving at the new apartment

Before the movers started bringing our belongings inside, they protected the new apartment.

Large protective sheets went down across the floors so that people could repeatedly walk in and out carrying furniture without scratching anything.

Even my shoes got their own protection.

This was another part of the service that reassured me.

There were several people moving large pieces of furniture around a completely empty apartment, but they were clearly taking precautions not just with our belongings, but with the apartment itself.

Then the process basically happened in reverse.

Boxes arrived, furniture came in, closets started filling up again, and what had been a completely empty apartment that morning slowly started looking like somewhere we actually lived.

Then my mother-in-law started putting red beans in the corners

At some point during all of this, I noticed my mother-in-law walking around the new apartment and putting dry red beans in the corners.

I had absolutely no idea what she was doing.

Apparently, this is connected to an old Korean 액막이 (aengmagi) tradition, customs meant to ward off bad luck and wish for peace and protection in a new home.

Red beans, or 팥 (pat), have traditionally been associated with 벽사 (byeoksa), the idea of driving away harmful or unwanted forces. Their red color was historically linked with vitality and positive energy, and red beans also appear in other Korean folk customs such as eating red-bean porridge around the winter solstice.

There are also moving-related customs involving putting small amounts of red beans or salt around the home, including in rooms or corners.

I found this Korean article explaining the practice, which describes salt as symbolizing purification and protection, while red beans are associated with warding off bad luck.

There doesn't seem to be one strict rule for how this is supposed to be done. The exact timing, amount and placement vary, and plenty of Korean families don't do it at all.

In our case, my mother-in-law simply placed a few dry red beans in the corners of the new apartment.

So apparently our new home now has a permanent +10 defense against bad luck buff.

What do you eat after moving?

Moving day also involved food, naturally.

My in-laws brought us some red-bean bread, which felt particularly appropriate given everything else happening that day.

There's another Korean food I had actually expected to encounter on moving day: 짜장면 (jajangmyeon).

Moving and Korean-Chinese delivery food have a pretty well-known association in Korea. The basic logic makes sense: your kitchen is packed away, the apartment is chaos, everyone is tired, and ordering something quick like jajangmyeon, jjamppong or tangsuyuk is considerably easier than cooking.

We didn't do that.

My in-laws decided we were having 족발 (jokbal) instead.

No complaints from me.

After spending the day watching our entire life travel from one apartment to another, sitting around our table in the new place and eating together was a pretty good way to finish the main part of moving day.

The move isn't quite finished when the boxes are unpacked

There is also an administrative part of moving in Korea.

As a registered foreign resident, changing apartments means you need to report your new place of stay, or 체류지 (cheryuji).

We did this by visiting our local 주민센터 (jumin center), the neighborhood community service center.

The official Korean government service for foreign residents is called 외국인 체류지변경신고 (oegugin cheryuji byeongyeong singo), or a report of change of place of stay.

Registered foreign residents are required to report the change within 15 days of moving into the new address. The current 15-day deadline has been in effect since December 10, 2020.

My wife, who is Korean, could have handled her address change online.

I tried to do mine online as a foreign resident, but the online process didn't work for me. Since we needed to go to the 주민센터 (jumin center) for mine anyway, we decided to just do both of our address changes there in person.

For me, the in-person process was straightforward.

I brought my Residence Card. I didn't need to show my passport.

The lease agreement itself is in my wife's name, so instead of using a lease with my own name on it, we used the document showing that my wife and I live together to prove that I was also moving into the new apartment.

So if the lease isn't in your own name, that doesn't necessarily mean you can't complete the address change. Depending on your situation, another document showing your relationship to the person on the lease and that you share the same household may work.

There was no fee for the change-of-address report.

You can check the current requirements on the official Government24 page for 외국인 체류지변경신고.

One terminology detail worth knowing is that you'll also hear 전입신고 (jeonip singo) constantly when moving in Korea. That's the standard move-in registration Korean residents deal with. For a registered foreigner, the procedure relevant to your immigration record is technically the 체류지 변경 신고 (cheryuji byeongyeong singo).

If you're moving as a Korean-foreign couple like us, the exact paperwork for each person can be slightly different, but we were able to handle both at the same 주민센터 (jumin center) visit.

So, what was moving in Korea actually like?

The part I expected to be the most annoying was the physical move.

Packing an entire apartment, moving expensive furniture, carrying everything across Seoul and then spending several days surrounded by boxes sounds exhausting.

In reality, 포장이사 (pojang-isa) made that part surprisingly painless.

The moving team packed our belongings carefully, protected our furniture, moved everything using the ladder truck at the old apartment, covered the floors of the new apartment and then unpacked and organized a surprising amount of it for us.

And they did the actual moving work in around five hours.

There was still plenty left for us to organize afterwards, especially those books, but we were already living in something resembling a home by the end of the day.

And around that very efficient moving operation were all these little details I probably wouldn't have thought about before living here:

the 사다리차 (sadaricha) outside the building, protective flooring, red beans quietly appearing in the corners, food with the family afterwards, and a 주민센터 (jumin center) form waiting for us once everything was over.

None of those things individually defines what moving in Korea is like.

But together, they made our moving day feel very Korean.

Now I just need to figure out where all those programming books are going.

Ready for your next move?
Visit Dev Korea to explore the latest job openings at dev-korea.com/jobs, or if you're hiring, post a job at dev-korea.com/post-a-job and connect with our growing international tech community.

Want to feel what the night was like? From the workshop to the networking and everything in between, here's a quick aftermovie of Dev Korea #12 🎥.

On August 11 we ran Dev Korea #12 at MARU. About 100 people came out.

This was our first proper workshop. Not a talk with questions at the end, but an hour where people spent most of the time practicing with each other. The topic was communicating clearly under pressure, aimed at interviews but useful for anyone who has to explain their work to a customer, a stakeholder, or a room.

Minson Vo told me upfront that he wanted people participating, not listening. I was nervous about it. Getting a room of developers to turn to a stranger and practice out loud is not a guaranteed win.

It went better than any of us expected. Everyone played along.

The premise

Minson has spent six years coaching engineers, analysts and PMs on communication, including at Microsoft and ThoughtWorks.

He opened with a friend back in Texas who had done more than 30 interviews in a few months. Every week a good one, every week a rejection a few days later. Not because he lacked skills. The point Minson built on it: most strong candidates don't lose interviews on ability, they lose because they over-explain, lose the thread under pressure, and don't sound like the senior person they are.

Tone and body language

The first half was delivery, and the single sharpest idea in it was inflection.

An upward inflection at the end of a sentence sounds uncertain and approval-seeking. A downward one sounds settled. He demonstrated the same self-introduction both ways and the difference was uncomfortable to hear. His target: land on a downward inflection about 80 percent of the time. Not always, or you sound grim.

On body language, the things research links to better interview scores are unglamorous. Eye contact. Nodding and reacting while the other person is still talking, rather than staring blankly to signal you're processing. Visible, moving hands.

He made a specific point about eye contact that landed with this room in particular. When people speak to someone they read as higher status or older, they look away more than the other person does. It's a deference habit, and it's stronger in Korean and other Asian cultures. It's also read as low confidence in an interview. Notice it and you've fixed one dimension for free.

Then everyone paired up. One person asked a question and practiced nodding and reacting; the other answered and practiced using their hands. Then they swapped. His disclaimer before starting: if this feels awkward, you're doing it right.

What came back afterward was more interesting than the exercise itself. One attendee said that nodding and smiling kept him more attentive, that he stopped drifting because his body was engaged in the conversation. Another described conversation as ping pong, where you're always anticipating what comes back.

Structure

The second half took on a belief that runs deep in technical fields: more detail is better.

His counter-argument was to put you in the interviewer's chair. Ten candidates between 10am and 3pm. You're tired. Someone walks in and talks for five minutes without pausing, in jargon, to impress you. They may well be smart. You don't remember them and you're exhausted. There's also a psychological cost to volume, since we associate more words with concealment.

Under pressure, most people speak in circles. Scripts evaporate. That's what fight-or-flight does.

His fix is having a structure in your head before you open your mouth. He gave three:

• PREP. Point, reason, example, point.
• Problem, pain, solution.
• Fact, opinion, reason. Useful for pushing back on someone tactfully.

He demoed PREP by answering "what's your favorite animal" twice, once meandering and once structured, and the structured version was noticeably easier to hold onto. The framing that stuck: structure sorts your words into compartments in the listener's head, so they can actually retrieve them later.

Second exercise, new partners, real interview questions from the handout. Answer using one of the structures, then get feedback from the person who asked, and write it down on your phone, because you won't remember it otherwise.

The self-diagnoses afterward were the best part of the night. One person realized he repeats himself and jumps around, 1-2-3 then 1-3-1-2-3. Another caught herself mid-answer thinking "what am I doing?" A third had done 30+ interviews and never used any structure at all, and had been assuming the interviewer already knew what he was talking about.

Q&A

The questions were unusually personal, which is a good sign.

Someone asked what to do when you're asked something you don't know. Don't invent an answer, and don't just say you don't know. Draw a boundary and then move to what you do know: I can't answer that exactly, but if you're asking about X, here's what I can tell you.

Someone asked what to do when you forget your words. Prevention: speak about 20 percent slower than feels natural, because nerves make you speed up. Recovery: pause, breathe, skip to the next thing, and mention what you missed at the end if it comes back.

Someone asked how to handle an interviewer with cold energy. Minson's answer came from a professor of his who stayed relentlessly positive with every student, and whose reason was that she went into every conversation looking for something to help the other person with. His own version is blunter, a phrase he repeats to himself: that ain't my energy.

And someone asked, honestly, how to feel confident around senior engineers when you feel less smart than them. He was careful with that one. He said it's a real problem and outside his expertise to solve psychologically, then offered the one thing he could: pick a small mechanical change, like the downward inflection, and use it next time you talk to them. The confidence follows the action, not the other way around.

His closing was a warning. The skills are useless without practice, and if nobody reflects after their next interview then the two hours were wasted. Five minutes alone afterward, write down what worked and what didn't, pick one thing, apply it next time.

A surprise five minutes

Then something we hadn't planned. Quincy Larson, founder of freeCodeCamp, happened to be in the room that night and got up for five minutes on how to use scaffolding to make your coding agents less dumb.

Unannounced, unrecorded, and the room was delighted. A decent argument for showing up in person.

The room

Food at check-in was pizza and burgers, the burgers again from Cry Cheeseburger.

The mood afterward was different from a normal event. People had spent an hour talking to strangers as part of the program, so networking didn't need a warm-up.

The event ended at nine. Everyone moved outside and kept talking in front of the building for the next two hours. I left at 11:15pm and there were still people out there, still going.

The feedback since has been the best of any event we've run. We'll do more like this.

Huge thanks

This event wouldn't have happened without:

• Minson Vo for a format we hadn't tried, and for showing us it works
• Quincy Larson for getting up with no notice
• MARU for the space
• Cry Cheeseburger for the burgers
• The volunteers who kept the night running: Bertijn, Brian, Nathaniel, Oscar, and Suhyun
• And everyone who paired up with a stranger and practiced out loud instead of sitting it out

Watch the workshop

The full session is online:

• Interview fearlessly: Speak confidently, stand out, and land a role faster, by Minson Vo

You can find every published Dev Korea talk on our talks page.

See you at the next one.

Ready for your next move?
Visit Dev Korea to explore the latest job openings at dev-korea.com/jobs, or if you're hiring, post a job at dev-korea.com/post-a-job and connect with our growing international tech community.

Want to feel what the night was like? From the talks to the networking and everything in between, here's a quick aftermovie of Dev Korea #11 🎥.

On July 29 we ran Dev Korea #11 together with AI Salon Seoul and SVC Seoul. More than 90 people came out. It was the first AI Salon chapter event in Korea, which made for a room heavier on founders than usual.

Two talks, both from people who had gone through the same thing from different angles: how do you get anyone to care about the thing you built.

Food was gimbap this time, served at check-in, which is where the first conversations always start.

Nobody owes your product attention

Cailyn Yong opened. She started 10X AI Club, an AI community in Korea that grew to 33K YouTube subscribers and an 800 member group chat, then founded Momo, a memory infrastructure product for AI agents. Her demo videos on X pulled 200K+ impressions each and brought in 600 users in a day. She raised her angel round off the back of them.

Her opening argument was blunt: anyone can build anything now. That means the product isn't the moat. The reason someone picks yours out of a dozen identical ones is you, which makes your story the defensible part.

The practical half of the talk was a teardown of her own demo videos, which she learned from a video producer at Founders, Inc.

• Keep it under a minute. Two minutes and nobody finishes it, including you.
• One second for a hook. Something visually odd enough to stop a scroll.
• Five seconds for a one-liner. Three or four words that explain the product.
• Then the problem statement, then straight into the demo. Don't explain what you built, show it.
• One feature per video. Three features means three videos, not one crowded one.
• End with the name in a font big enough to be annoying.

Her framing for the demo itself was the part people wrote down: play your ICP. In one video she acted as a CEO who couldn't keep track of what her team was doing, and walked through the product from inside that problem. It's more convincing than describing the problem from the outside, because it proves you understand it.

On going viral, she was direct that it isn't luck and it isn't one post. Warm up the account. Post whether or not anyone is watching. Then study the accounts already winning in your space. In memory infrastructure that meant Supermemory and Hypernotes: she read through what they'd posted, worked out which ones landed, and reused the templates. Eventually one of yours breaks through, you won't know which, and that's your first thousand followers.

She also drew a line under rage bait. She'd seen it work and told the room not to do it, because people remember what kind of person you were on the way up.

One last observation that surprised the room: X brings volume, YouTube brings quality. Her highest-retention users came from long-form video, because sitting through twenty minutes is a filter.

Seven pivots

Jun Kim followed with the story of getting to Aside, an AI browser that works across sites you're already logged into. It launched a month before the event and did 2.5 million views and 20K users in its first week.

Then he put up the slide that set the tone: six YC rejections before the seventh application got in.

The tour through the pivots was honest in a way these talks usually aren't. An AI executive assistant they built for months and killed without ever launching, because they lost conviction. An AI project manager that fell over in San Francisco partly because "project manager" doesn't mean one specific job there the way it does in Korea. An AI coding agent for SDK instrumentation. An error investigation tool that got them US user interviews and no customers.

Then Caret, a note taker with live translation, built by his co-founder over his objection that the space was too crowded. It made $2,000, which was more than the company had earned in all of 2024. It went viral in Korea. And it got them a YC interview that ended on one question: where's your revenue from? Ninety percent Korea. That was the rejection.

The three lessons he closed on:

Look for overlooked problems. He asked the room what problem Spotify solved in 2006. Someone said music streaming. His answer was that downloading music over torrents was slow and full of viruses. Good problems look too small to be worth solving right up until you notice they're the bottleneck.

Users don't know what they want. His fix was shortening the feedback loop rather than running more interviews. On Aside the team is its own ICP, which breaks a startup rule but takes iteration from weeks to hours. For B2B, he argued your first five design partners set your product direction, so getting five Korean enterprises when you want to sell in the US is worse than useless.

Conviction matters more than ARR. They found something that looked like product-market fit in sales assistance and lost their energy anyway, because they didn't care about the customer. His two questions: is this a customer you want to serve, and would you be happy doing this even if it fails?

Every pivot had an exit condition set in advance. Ten active customers in three months. $50,000 ARR. Ship in two weeks or stop. That's what let them move fast without it feeling like giving up each time.

Q&A

The Q&A ran until we had to cut it for networking, which is the good version of running out of time.

Somebody asked about naming, given that the company had cycled through Append, Ace, Caret and Aside. Jun's rule was that names start with A, for alphabetical ordering, and that the real constraint was always the price of the .com.

Someone else asked how they survived two years financially. Three co-founders and one teammate, no salaries, and the company account nearly empty before YC. Staying small is what bought the runway.

There were good questions on security, on whether people will still use browsers in thirty years, and on whether acquisition is the exit plan. That last one got the best laugh of the night: nobody had reached out yet, but he hoped they would.

The room

The Q&A ran so long we had to drop the open mic, which felt like the right trade. People pitched to each other during networking instead.

Huge thanks

This event wouldn't have happened without:

• Cailyn Yong and Jun Kim for two talks that were unusually specific about what didn't work
• Jay Kim and AI Salon Seoul for bringing the first chapter event to Korea
• Lucy and SVC Seoul for hosting us
• The volunteers who kept the night running: Bertijn, Brian, Nathaniel, Oscar, Serin, and Sonali
• And everyone who showed up, asked questions, and stayed to talk

Watch the talks

Both talks are online:

• How to go viral on X and build your community, by Cailyn Yong
• 7 pivots to build something people want, by Jun Kim

You can find every published Dev Korea talk on our talks page.

See you at the next one.

Ready for your next move?
Visit Dev Korea to explore the latest job openings at dev-korea.com/jobs, or if you're hiring, post a job at dev-korea.com/post-a-job and connect with our growing international tech community.

Want to feel what the night was like? From the talks to the networking and everything in between, here's a quick aftermovie of the OpenAI Codex Meetup 🎥.

On July 8, together with OpenAI and Hashed, we ran the OpenAI Codex Meetup in Seoul. Our tenth Dev Korea event.

Demand for this one broke every record we had. More than 2,500 people RSVP'd. The room holds a fraction of that, so we could only approve a small share of them, and turning away that many people is not a fun position to be in.

More than 250 people made it in. That makes it our biggest event ever by some distance, and the room felt it. ICML was in town that week, so alongside the usual Seoul crowd there were a lot of people who had flown in and were looking for somewhere to meet other builders.

The evening

Gabriel Chua opened with "Valuemaxxing with Codex", on getting real leverage out of the agent rather than treating it as autocomplete.

The timing was good. The GPT-5.6 family, Sol, Terra and Luna, had been in limited preview since late June, and general availability landed the day after our event. So the room was hearing about it roughly 24 hours ahead of the public launch, which is not something we can usually promise.

Won Park then joined Gabriel on stage for a fireside Q&A, and the questions kept coming until we had to cut it off for the break.

Community demos

After the break, five people from the community showed what they had been building with Codex. This is always the best argument for showing up in person.

Hamin Lee

Jaeyun Ha

Jineui Kim

Byungjun Kim

Neel Somani

The room

Food was a buffet this time rather than the usual pizza, and it went the way food goes at a 250 person event.

The OpenAI swag was the other story of the night. Water bottles and caps. The caps were gone almost immediately, which surprised nobody except us.

One note: the talks weren't recorded this time, so there's no video to link. If you missed it, this recap is what we have.

Huge thanks

This event wouldn't have happened without:

• OpenAI for bringing the team, the updates, and the swag
• Hashed for hosting 250+ people in their space
• Gabriel Chua and Won Park for the talk and the fireside
• Hamin Lee, Jaeyun Ha, Jineui Kim, Byungjun Kim, and Neel Somani for demoing what they had built
• The eight volunteers who kept a 250 person check-in moving: Aleksey, Bertijn, Brian, Nathaniel, Oscar, Rob, Sonali, and Suhyun
• And everyone who showed up and made it the biggest one yet

If you were one of the 2,000+ who didn't get a spot, sorry. We're working on it. The best way to hear about the next one early is our newsletter and Discord.

You can find every published Dev Korea talk on our talks page.

See you at the next one.

Ready for your next move?
Visit Dev Korea to explore the latest job openings at dev-korea.com/jobs, or if you're hiring, post a job at dev-korea.com/post-a-job and connect with our growing international tech community.

But at some point you want to get out of the city. A weekend in the countryside, a trip somewhere the KTX doesn't reach, a drive along the coast. And that's when you discover that a French license in your wallet does nothing for you here. I'd never bothered getting an International Driving Permit before leaving France, and even if I had, an IDP only covers your first year in Korea anyway, so mine would have run out long ago. To drive here as a resident, you need a Korean license.

The good news is that around 130 countries appear on KOROAD’s acknowledged list. If yours is one of them, you exchange your license for a Korean one instead of testing for it: no road test, no driving school, no 500,000 KRW package. Most acknowledged countries are also exempt from the written exam, though not all of them, so check yours before assuming. More on that at the end.

The less good news is that the paperwork around it is where people get stuck. Not the license itself. The certificate you didn't know existed, the embassy that stopped issuing the document KOROAD asks for, the photo size.

I just picked up my Korean license, so while the whole thing is still fresh in my head, I wanted to write down exactly what it took. Hopefully it saves someone else a wasted trip.

First: does this apply to you?

This article covers the exchange route (면허 교환), not the full test route. You can use it if:

• You hold a valid, full foreign driving license (learner, provisional, probationary and temporary permits don't count)
• Your license was issued by a country Korea has a mutual recognition agreement with
• You're a registered foreign resident in Korea (if you've been here under 90 days without an ARC, you can't exchange)

A few limits worth knowing up front:

• Only car licenses are exchanged. Motorcycle and truck categories are not transferred. Those require taking the Korean practical exam.
• You'll typically receive a Class 2 ordinary license (2종 보통), which is fine for a normal car but won't cover vehicles like a 9+ seater van (Carnival, Starex).
• Your foreign license is surrendered. You hand it over at the counter and sign a consent form agreeing to it. You can get it back, but there are conditions. More on that below.

If your nationality is different from the country that issued your license, there's an extra condition: you have to prove you stayed in that country for more than 90 days, using school or employment records, a tax invoice, a permanent residence card, citizenship, or a long-term visa of over a year.

The documents

This is the part that determines whether you walk out with a license or walk out with nothing. The Gangnam office has the full list posted at the counter, and it's worth photographing when you arrive:

1. Original foreign driver's license, valid and not expired
2. Original passport, showing the stamp of your most recent entry into Korea
3. Original Alien Registration Card (or Domicile Registration Card)
4. Original Entry & Exit Certificate, reference period from your birthdate to a recent date. Valid for 2 months.
5. Original Embassy Certificate or Apostille about your license. Valid for 1 year from its issue date.
6. Three color photos, 3.5cm × 4.5cm, passport style, white background
7. Processing fee

Two of those validity windows matter a great deal for planning, and I've not seen them mentioned in any online guide:

• The entry/exit certificate expires after 2 months. Don't get it early.
• The embassy certificate is good for a year. So the correct order is: sort the embassy paperwork first, then get the entry/exit certificate shortly before your visit.

The notice also spells out what an acceptable embassy certificate or apostille looks like, and the requirements are stricter than you'd expect. It must be an official original from your consulate or country representative agency, written in English or Korean. If it's in another language, it has to be translated and notarised. It must state the issue date, expiration date, license number, license type, your name and your date of birth. For apostilles specifically, a copy of both sides of the license must be included, and if the apostille has been damaged or detached it's treated as invalid. Don't unstaple anything.

Let me break down the two that actually cause problems.

1. The embassy certificate (or apostille)

Korea needs proof that your license is genuine. Depending on your country, that means either a certificate from your embassy in Seoul, or an apostille from your home country.

For French citizens, this is a two-stage process, and the first stage is the one that catches people out.

What the Korean authorities want is a document certified by the embassy confirming that your license is valid and that the driving entitlements attached to it are current. To issue that, the embassy needs your ADCS, the attestation de droits à conduire sécurisée (also called the relevé d'information restreint). You request the ADCS yourself from France, via service-public.gouv.fr. Nothing at the embassy can start until you have it in hand.

The age of your license doesn't matter here. Mine dates from 2013 and had sat unused in a drawer for years. What matters is that your entitlements are still live, and the ADCS is precisely the document that proves it.

Once you do, the certificate request goes to the consular section in Seoul by post. There's no counter appointment for this. You send:

• The original of your French license. Not a copy. It goes in the envelope, and it comes back to you together with the certificate.
• Your ADCS
• A copy of your passport
• The chancellery fee, paid by postal money order (통상환증서)
• A stamped, self-addressed envelope for the reply

Posting your original license is uncomfortable, and it's worth knowing in advance so you're not caught off guard. If needed, use a tracked service for the outbound envelope, and be aware you'll be without your license for the duration of the round trip.

The request itself can be written on plain paper. There's no form to fill in.

The fee: 69,720 KRW if you're registered on the consular register (inscrit au registre des Français établis hors de France), 139,450 KRW if you're not. I was registered, so I paid the lower rate. That's a 70,000 KRW gap for something that costs nothing to sort out, so if you're a French resident in Korea and haven't registered with the consulate, do it before you send this request. Note that these amounts can change on the 1st of any month, so check the embassy page before you buy the money order.

Postal address:

Ambassade de France en Corée (permis de conduire) 43-12 Seosomun-ro, Seodaemun-gu, Séoul 03741 주한프랑스대사관(운전면허증) 서울특별시 서대문구 서소문로 43-12 (우) 03741

So the realistic sequence for a French license is: request the ADCS from France → post everything to the embassy → wait for the certified document to come back by mail → then book your morning at the license center.

A critical warning for other nationalities. Several embassies in Korea have stopped issuing license confirmation documents, and applicants from those countries are directed to submit an apostille instead. As of writing, KOROAD's guidance names the American, British and Finnish embassies among them, and the British government has confirmed its embassy no longer issues validation letters.

If that's you, here's the part that hurts: apostilles cannot be processed inside Korea. You have to arrange it in the country that issued your license, which usually means enlisting someone back home or using a paid service. This is the step most likely to add weeks to your timeline, so start it before anything else.

Which embassies are affected changes over time, and I'm only in a position to vouch for the French process. So before you do anything, check two things: KOROAD's current guidance, and your own embassy directly. KOROAD explicitly advises that if your embassy hasn't published a notice suspending the service, you should contact them in advance to confirm they can still issue a certificate.

2. The Certificate of Entry & Departure (출입국사실증명서)

It's a record of every time you've entered and left Korea, and KOROAD needs it to verify that you were actually living abroad when your license was issued.

I didn't have it when I showed up. The staff handed me a printed sheet with the two nearest community service centers on it and sent me off.

Two things that make this much less painful than it sounds:

• The reference period must be from your date of birth to today. Not "the last five years." If you request the wrong range, you'll be sent back.
• You get it at any 주민센터 (community service center), not at immigration. It took a few minutes at the counter and cost 2,000 KRW. If the nearby centers are closed, an immigration office can also issue it.

For the Gangnam Driver's License Examination Office specifically, the two closest are Samseong 1-dong Community Service Center (봉은사로 616) and Daechi 2-dong Community Service Center (영동대로 65길 24).

Honestly, if I were doing it again I'd just get this before going. But if you forget, it's a 30-minute detour, not a lost day.

Step by step at the license center

I went to the Gangnam Driver's License Examination Office (강남운전면허시험장), chosen on the strength of its reputation for being the most English-friendly center in Seoul. That held up. The signage at the foreign license counter is bilingual, the forms come in English and Korean, and the staff deal with foreign applicants constantly enough that nothing about my paperwork surprised them. If your Korean isn't strong, this is the one to go to.

Seoul also has a Seobu location, and Busan, Incheon and Daegu centers handle foreign applicants. Not every branch runs the same way, so it's worth calling ahead.

Posted hours for license exchange are 9am to 4pm. In practice that seems to be the cutoff for starting an application rather than a hard closing time. I was still being processed at around 4:30pm and the staff were working away quite happily. But I wouldn't rely on that: the process spans three counters on three floors, and if the eye test window or the exchange desk has stopped taking new applicants you're coming back another day. Go in the morning and give yourself room.

Here's the actual flow, which a Dev Korea community member mapped out for me before I went, and which turned out to be accurate end to end:

1. Photos (2nd floor). If you don't already have photos that meet the spec, there's a booth in the building. Six photos for 10,000 KRW. Skip this step if you brought your own.

2. Foreign license counter (1st floor, right inside the entrance). There's a dedicated line for foreign license applications. In practice there's rarely a queue. Take a number, hand over your documents. The officer checks everything, tells you what's missing or what to sign, keeps the file, and sends you for the eye test.

3. Eye test (basement). Pay 6,000 KRW at the window. Stand at the desk, cover one eye at a time, read the lines the examiner points at. Genuinely a 15-second affair. They hand you a sheet with the result.

4. Pay the issuance fee (back on the 1st floor). 10,000 KRW for the standard physical card, or 15,000 KRW if you also want the mobile ID version, the same system as the mobile ARC. I paid the 15,000 and would recommend it; the physical card is what you scan to activate the digital one.
5. Wait for your name (1st floor, opposite side near the water cooler). Names come up on a large monitor. You collect the physical card there.

If you paid for the mobile version, the setup happens afterwards on your phone. Download the 모바일 신분증 app, verify your identity and set a PIN, then scan the IC chip in your new physical card to register it. Korea's mobile driver's license can also be carried in the Naver app now, and the mobile ID carries the same legal force as the plastic card.

Then you're done. You walk out with a Korean driving license the same day.

What happens to your foreign license

At the counter you sign a Foreign License Exchange & Termination Agreement (외국면허 교환 및 폐기 동의서). It's a single page asking for your name, ARC number, issuing country, foreign license number, issue and expiry dates, whether you hold or have previously held a Korean license, and your purpose of visit to Korea. You sign it under penalty of perjury.

Read the note section before you sign, because this is the part people misunderstand. Your surrendered license is held for 10 years and then destroyed if nobody claims it. Within that window, there are exactly three ways to get it back:

1. You request its return when travelling abroad. You need to go in person, show your Korean driver's license, and hand in a confirmed, printed flight ticket.
2. The authorities of the issuing country request it.
3. You cancel your Korean driver's license, voluntarily giving up the license you just exchanged for.

A poster at the counter puts it more bluntly than the form does:

Three things it makes explicit:

• You surrender the license once, not every time. The Korean office describes this as a temporary submission.
• Nobody else can collect it for you. 본인만 가능, in person only.
• Going abroad is the only accepted reason. You cannot retrieve it to renew it back home, or for any other purpose.

That last point is the one to sit with. If your French license is approaching expiry and you were assuming you could fetch it, renew it in France and carry on, that's not on offer. It comes out of the drawer when you have a boarding pass, and not otherwise.

Note also that this is departure abroad, not emigration. You don't have to be leaving Korea forever, but you do need a confirmed ticket in hand, which rules out casually collecting it on a quiet afternoon.

One more detail that's easy to miss: your license can only be returned at the office where you exchanged it. If you did the exchange in Gangnam and later move to Busan, that's a trip back to Seoul.

In practice, for most people this means accepting that the Korean card is now your license for the duration of your stay. If you go home once a year and want to drive there, plan around it rather than assuming you can pick your old license up on the way to the airport.

A reassuring note for French citizens specifically. If you eventually move back to France for good, you don't actually need the physical card sitting in the drawer in Gangnam. Because you originally passed your test in France, you can apply for rétablissement de vos droits à conduire within the year following your resettlement, and your Korean license is recognised in France during that window. If you'd passed your test in Korea instead, you'd exchange the Korean license for a French one, which requires at least 6 months of residence in Korea.

One thing to handle before you leave Korea, though: that process needs a French translation of your Korean license and of your driving record certificate (운전경력증명서). The record certificate has to be collected in person at a police station in Korea, and the French Embassy has been explicit that they cannot send it to you afterwards. Pick it up while you're still here.

What it actually cost

Under 35,000 KRW at the center itself. The embassy certificate is by far the largest line item, and it more than doubles if you're not on the consular register. Amounts can change monthly, so confirm before sending your money order. If you're not French, check your own embassy's current fee, or the apostille cost from your home country.

One note on the official numbers: the notice at the counter quotes a processing fee of 16,000 KRW without the written test, or 26,000 KRW with it. That 16,000 is the 6,000 eye test plus the 10,000 standard card. I paid 21,000 because I opted for the 15,000 mobile version instead of the 10,000 plain card. So if your country is exempt from the written test, budget 16,000 to 21,000 at the center, plus photos if you need them.

Things I'd tell someone starting this today

• Start with the embassy certificate or apostille, and start it early. Everything else can be sorted in a single morning. This one has real lead time. For French citizens it's an ADCS from France plus a postal round trip with the embassy. If your embassy no longer issues certificates, it's an apostille arranged from your home country, which is slower again.
• French citizens: register on the consular register. It halves the certificate fee, from 139,450 to 69,720 KRW.
• Bring originals of everything. Passport, ARC, license. Copies won't do.
• Get the entry/exit certificate before you go, with the reference period set from your birth date. 2,000 KRW and five minutes at any 주민센터.
• You cannot send someone else. Proxy applications aren't accepted, so you have to show up in person.
• Mind the two expiry windows. The embassy certificate or apostille is valid for 1 year from its issue date. The entry/exit certificate is valid for only 2 months. Get the embassy document first, the entry/exit certificate last.
• Go in the morning. The exchange counter runs 9am to 4pm.
• Check your passport has your most recent entry stamp, since the notice specifically asks for it.
• Accept that you're giving up your home license. It's recoverable, but only with an outbound flight ticket, only in person, and only at the office where you exchanged it. For the duration of your stay in Korea, the Korean card is your license.

Your country may work differently

I want to be clear that this is one nationality's experience. France has a straightforward mutual recognition agreement and an embassy in Seoul that still issues license certificates, which puts me in the easier category, even if that certificate costs more and takes longer than anything else in the process.

Beyond that, the variation is real and I'm not going to pretend to summarise it. Some countries exchange for an equivalent license type rather than the default class. Some appear on the acknowledged-countries list but still require the written exam, including certain US states. Which embassies issue certificates changes. Any list I published today would be quietly wrong within a year, and you'd be the one making a wasted trip on the strength of it.

So for your own case, go to the source: KOROAD's English guidance on license exchange for the acknowledged-countries list and document rules, your embassy for the certificate question, and the license center itself for anything the first two leave ambiguous. KOROAD advises contacting the regional office in advance regardless, since the exchange takes time to process.

Final thoughts

The license center staff deal with foreign applicants constantly, the foreign license lane is quick, and the whole on-site portion is a single morning.

The difficulty is entirely front-loaded into two documents: the embassy certificate and the entry/exit record. Sort those out first and the rest is a formality.

Note: KOROAD requirements, acknowledged-country lists, embassy services and fees change. Verify current requirements with KOROAD's English license exchange guidance or your nearest Driver's License Examination Office before you go. This article is for informational purposes only and is not legal advice.

Ready for your next move?
Visit Dev Korea to explore the latest job openings at dev-korea.com/jobs, or if you're hiring, post a job at dev-korea.com/post-a-job and connect with our growing international tech community.

Want to feel what the night was like? From the talk to the networking and everything in between, here's a quick aftermovie of Dev Korea #9 🎥.

On June 17 we ran Dev Korea #9. More than 110 people came out for the evening. Developers, platform and infra folks, founders, a good mix from across the Seoul tech scene, all there for one long talk and the usual tail of conversations afterward.

Jan de Vries, a consultant, trainer and coach based in the Netherlands, and the founder of Antifragility.works and BlueOceanRecon.com, was in Korea for two weeks, his first time in the country, and decided to spend one of those evenings giving a talk with us. He has been doing antifragility talks at conferences in Belgium, Spain, Switzerland, Luxembourg, Czech Republic, Australia, Japan, Malaysia and Canada over the last three years. Seoul is now on that list.

He opened with a show of hands. Four people in the room were familiar with antifragility. Nobody was working in risk management. By the end of the hour it was clear why he asked both questions together.

Risk management is risk theater

Jan started with a consultancy report about coping with unknown risks. Read the actual strategies in it, he said, and almost every one is about known risks. You end up preparing for the failures you already have names for, and not for the ones you don't.

That's the case Nassim Nicholas Taleb makes when he calls risk management risk theater. His alternative is not better prediction. It's becoming less fragile to risk in the first place, and ideally benefiting from it.

Until 2012 we only had two words for something that isn't fragile: robust and resilient. Antifragile added a third. Jan walked through all four with a picture for each.

• Fragile is glass. Add stress and it breaks.
• Robust is reinforced concrete, or a presidential limo. It holds for a long time, then it breaks anyway.
• Resilient is the Phoenix. It comes back, but it comes back as the same bird. In software, that's an Airbus A380 cockpit: hydraulics fail, backups take over, it keeps flying, and it is never better than it was designed to be.
• Antifragile is the Hydra. Cut off one head and two grow back. Every attack makes it stronger.

The IT version of the Hydra is Netflix's Chaos Monkey, which kills things in production between 9am and 3pm so that engineers fix them during office hours instead of at 2am. There was a latency monkey injecting artificial delay, and another one hunting down unused cloud resources and switching them off, with a 30 minute grace period before it came back around. Netflix eventually became immune to the Chaos Monkey, so they built the Chaos Gorilla to break bigger things. The whole Simian Army is on GitHub if you want to point it at your own systems. Test environment first.

He also made a point about layers. A single aircraft is resilient. The aviation industry is antifragile, because it learns from every crash. When the 737 MAX went down in 2018 and 2019, every similar aircraft in the world was grounded until the cause was understood.

Technical debt is the most fragile thing in IT

He asked the room what the single most fragile thing in software is. The answer he was looking for was technical debt.

His graph for it is the one worth stealing. The cost of change always rises as software ages, and that's fine, that's the optimal line. Technical debt is the gap between that line and what your changes actually cost. It gets dangerous when the curve turns into a hockey stick, because past that point the software is beyond repair and your responsiveness to customers is gone. He quoted Gene Kim's definition, which is still the best one: technical debt is what you feel the next time you want to make a change.

Getting rid of it is what Taleb calls via negativa, the power of subtraction. Our instinct is to add: another process, another layer, another tool. Via negativa says remove the feature, remove the dependency, remove the layer.

The practical version he showed came from Project to Product: a four-color backlog. Green is features, red is defects, blue is paying down debt, orange is mitigating risk. Customers only ever ask for green and red. Teams that only deliver green and red watch their defect count climb forever. The teams that keep spending on blue and orange are the ones whose defect count eventually comes down. He suggested a layered kanban board as a simple way to hold yourself to that split.

He then asked the room which balance matched their own workplace. Somebody answered "a column with no green at all," which got a laugh. There's a word for that, Jan said. It's called a refactoring sprint.

Optionality, and why vendor lock-in hurts

The middle of the talk was built around three forms of asymmetry. The first was nonlinearity: anything with more upside than downside from random events is antifragile, and the reverse is fragile. The trick to spotting it is asking whether the thing accelerates toward pain or toward gain. Jan's memory aid is that the fragile curve is a frown and the antifragile curve is a smile.

Deployments per month is the classic smile. Going from zero to a few deployments hurts a bit. Keep pushing and you end up with a working CI/CD pipeline, and at that point shipping ten times a day or three times a month stops mattering. Continuous deployment is antifragile. Coordinated tasks inside a software project is the frown: coordination helps at first, then dependencies and deadlines pile up and it accelerates toward pain. Software projects are fragile.

The second was optionality. When people hear the word "options" they think financial options, which are expensive. Most of the useful options are non-financial, and cheap or free, and we just don't recognize them as options.

The clearest anti-pattern is vendor lock-in. He put an ERP strategy next to a best-of-breed strategy and worked through the criteria. Almost every advantage lands on the best-of-breed side. Even integration, which is supposed to be the ERP selling point, usually isn't as good as promised, while a standalone vendor treats integration as a knockout criterion because they don't get bought otherwise. His conclusion: ERP systems are fragilizers.

Other ways to buy yourself options, most of them cheap:

• T-shaped profiles, which give the individual more moves and give the employer cross-pollination and fewer silos
• A bus factor of at least three. A bus factor of one means one resignation ends the project
• Real redundancy. His analogy: if you always sleep on your left side you technically only need half a mattress, but that's zero redundancy, and the first thing that goes wrong ruins your night
• Observability. Monitoring covers the known unknowns. Observability is what you have for the unknown unknowns, which is exactly the territory antifragility cares about
• The barbell strategy. Put 80 to 90 percent of your resources into boring, safe, predictable work, and 10 to 20 percent into genuinely speculative bets. Avoid the middle
• Microservices, which turn one big irreversible decision into many small reversible ones. If one turns out badly you rewrite it instead of rewriting the empire

That last one comes with a line that stuck with a lot of people: software has diseconomies of scale. Milk is cheaper in bigger cartons. Software is cheaper in lots of small ones.

He also asked the room whether MTBF or MTTR is the antifragile metric. MTTR, and it isn't close. There are two cases where you optimize mean time between failures instead: hardware you launched into space, and a medical device implanted in someone's chest. Everywhere else, optimize for recovery, because antifragile systems assume failure will happen. They're supposed to break a little, all the time. That's the part that makes them stronger.

Skin in the game

The third asymmetry was fragility transfer: one party takes the downside while another takes the upside. The reverse of that is skin in the game, meaning you have something real to lose.

His example goes back 3,700 years. Hammurabi's code 229 said that if a builder builds a house and the house collapses and kills the owner, the builder is put to death. Houses in Mesopotamia got noticeably more stable after that.

Then he pointed the same lens at how we organize software work:

• Silos are fragilizers. A dev in a dev silo has no skin in the ops game and vice versa. This is the actual reason DevOps works: what I build in the morning I maintain in the afternoon.
• Projects are fragilizers. A project manager owns the delivery and the initial cost, and usually owns none of the operating cost that follows for the next twenty years.
• Command and control is a fragilizer. Information travels up to authority, a decision comes back down, and everything is slow and worse for it. The fix is to push authority down to where the information already is, which needs technical competence, organizational clarity and shared priorities to work.

He closed this section with organizational debt, which Steve Blank describes as technical debt but worse. It's what you have when your hierarchy actively works against your operational flow. And it isn't a fair fight, because Conway's law says your systems will end up mirroring your communication structure whether you like it or not. Published in 1967, still unbroken. The upside is that you can also use it deliberately: simplify the organization if you want simpler systems.

His sharpest line here was about Frederick Taylor, who published his five principles of scientific management in 1911 and died in 1915. Management decides, workers execute. If that still describes your company, you're running a 115 year old operating model built for assembly lines, on software.

Most of the reasons an organization isn't ready for digital transformation aren't technical, he said. Technically it can be solved. Organizationally it stays a problem.

Don't be a fragilista

The last slides were his advice, and it's counterintuitive on purpose.

A CI/CD engineer who suppresses every small failure to keep the pipeline green is setting up a much larger failure later. You don't want small failures hidden. You want them exposed, visible, and known to everyone. Spend your error budget. Ship the high-risk high-reward feature. Upgrade the old infrastructure. Run the disaster recovery drill.

Avoiding black swans is impossible. Building something that survives them isn't.

A full room and a long Q&A

More than 110 people stayed through the evening. Doors opened at 6:30 and the pizza and the burgers went fast during check-in, which is usually where the first round of conversations starts before anyone sits down.

The Q&A ran long, in a good way. It started during the talk and never really stopped. People asked about infrastructure as code and avoiding cloud lock-in, whether AI-assisted development creates technical debt faster than it clears it, how to make a case for antifragility inside a risk-averse company, what to read on the organizational side, and whether antifragility is worth the effort for every team or only some.

A great moment happened between two attendees. Someone asked a distributed infrastructure question Jan said was past his technical depth, and someone else in the room picked it up and answered from their own experience building anomaly detection. Jan's follow-up was whether those anomalies were made visible across the organization. They were. That's the point, he said. You should be proud of the things you found that went wrong.

And yes, someone asked him if he knew there's a K-pop song called Antifragile. He didn't, but he did know that his Google alert for the word is mostly music.

Plenty of people stuck around for the networking hour afterward anyway, food or no food.

Huge thanks

This event wouldn't have happened without:

• Jan de Vries for spending one of his two weeks in Korea giving us an hour of this, and for being honest about the parts he doesn't have answers for yet
• Cry Cheeseburger for the burgers
• The volunteers who kept the night running: Sonali, Brian, Chanwu, Oscar, and Rob
• And everyone who showed up, asked questions, answered each other's questions, and stuck around to talk

Watch the talk

The full talk is online:

• Beyond resilience: Architecting antifragile systems, by Jan de Vries

You can find every published Dev Korea talk on our talks page.

Ready for your next move?
Visit Dev Korea to explore the latest job openings at dev-korea.com/jobs, or if you're hiring, post a job at dev-korea.com/post-a-job and connect with our growing international tech community.

Want to feel what the night was like? From the talks to the networking and everything in between, here's a quick trailer of Dev Korea #8 🎥.

On May 18 we ran Dev Korea #8 with Salmon. Around 150 people came out for the evening. Developers, QA engineers, a good mix of folks from across the Seoul tech scene, all there for two talks and the usual long tail of conversations afterward.

Salmon made the night happen. They came on as the main sponsor, brought their speakers, handed out swag, and even set up a photobooth that people kept coming back to all evening.

Both talks were about software quality, but they came at it from opposite ends. One was about how we write failures into our code. The other was about how teams decide what's actually worth testing before a release. Put together, they were really asking the same thing: how do we build software that's easier to reason about?

Making failures easier to understand

The first talk was by Sergey Chernov, Lead Software Engineer at Salmon Group Ltd, on domain errors and functional error handling in Kotlin.

He started with a problem most backend developers will recognize: a method signature rarely tells you what can actually go wrong inside it. A function looks simple from the outside, but the implementation might be full of expected business failures. An invalid code, an expired document, a signing window that already closed, a document that was already signed, a policy rule that rejects the request.

When those failures live inside the implementation, the only way to know they exist is to open the method and read it. That's annoying with your own code and genuinely painful with old code, unfamiliar code, or code an AI agent wrote.

Sergey's argument was that expected failures belong in the method contract, not buried in the body. Instead of leaning on runtime exceptions, he showed how to make success and failure explicit in Kotlin using sealed interfaces, typed result models, and the Either approach.

He also pushed back on the instinct to create one big shared error type. He'd rather define a narrow union of errors per method, so each method only advertises the failures it can really produce. The caller stops handling impossible cases, and the compiler starts working for you instead of against you.

The cost is real: more types, more mapping code, more verbose signatures. What you get back is clearer contracts and error handling you can actually predict. As he put it, the goal isn't to get rid of exceptions. It's to stop hiding the failures you already know about.

From more tests to better focus

The second talk was by Artemii Zakharov, Lead QA Engineer at Salmon Group Ltd, on moving from test cases to risk management.

His starting point was very concrete: regression testing stops scaling at some point. His team had around 2,000 test cases, roughly 500 automated and the rest manual. With eight QA engineers, that was about 57 hours of regression on paper, and once you added analysis, retesting, back-and-forth, and bug fixes, a release could take around five days just to validate.

The issue wasn't a shortage of test cases. It was that piling on more of them had stopped helping. As a product grows and its parts get more connected, one small change can ripple into onboarding, payments, support, conversion, or real money moving around. More tests doesn't mean more attention on the things that actually matter.

That's the case for risk-based testing. It isn't about testing less at random. It's about figuring out which areas can hurt the business most and putting your effort there. Artemii framed it around three questions: how much would this hurt users or the business (impact), how far can the change reach (change surface), and how much protection already exists, like automated tests, reviews, or monitoring (safety net).

The part that got the most interest was how his team uses AI for release impact analysis. The AI doesn't make the call. It produces a readable report flagging where the team should look, sorting test cases into must test, recommended, and can skip. That last bucket is the valuable one. Skipping a low-risk area isn't carelessness; it's saving human attention for the parts of the release that can actually do damage.

He was honest about the limits too. AI won't catch runtime dependencies, hidden blast radius, weak test mapping, brand-new features, or business-critical boundaries unless the team feeds it that context. The point wasn't that AI replaces QA. It's that AI can take the repetitive analysis off your plate so QA engineers can spend their time on judgment, exploration, and the business context only they have.

A full room and a lot of Salmon energy

Around 150 people stayed through the evening. They came for the talks but hung around to meet each other, ask questions, and take photos.

The burgers and pizza went fast. They kept the room fed between talks and gave people a reason to stand around and chat instead of rushing off.

The Q&A after both talks was genuinely good. People asked about Kotlin vs Java, how to keep error types from sprawling, how to enforce typed error handling across a big team, what AI-assisted QA looks like in a real repo, and how teams deal with human error, non-technical bug reports, and test coverage after a release ships.

The photobooth was a nice touch, a casual way to grab a few photos that made the night feel more like people hanging out than a conference. That's the kind of evening we're trying to build: somewhere you can learn something and also feel part of Korea's international tech scene.

Huge thanks

This event wouldn't have happened without:

• Sergey Chernov and Artemii Zakharov for two honest, practical talks
• Salmon Group Ltd for backing the event
• Maru by the Asan Nanum Foundation for hosting us
• Cry Cheeseburger for the burgers
• Everyone who helped run the night: Aleksey, Bertijn, Brian, Marcus, Oscar, Sonali, Suhyun, Svetlana, and Valentina
• And everyone who showed up, asked questions, took photos, shared posts, and stuck around to talk

We're looking forward to the next one.

Watch the talks

Both talks are online:

• Domain errors and functional error handling in Kotlin, by Sergey Chernov
• From test cases to risk management, by Artemii Zakharov

You can find every published Dev Korea talk on our talks page.

Ready for your next move?
Visit Dev Korea to explore the latest job openings at dev-korea.com/jobs, or if you're hiring, post a job at dev-korea.com/post-a-job and connect with our growing international tech community.

Want to feel what the night was like? From the talks to the networking and everything in between, here's a quick trailer of Design Korea #0 🎥.

On April 23, we hosted Design Korea #0, our very first design-focused Dev Korea event, bringing together 110+ designers, developers, founders, and builders in Seoul.

Made possible with the support of Figma, the event focused on a simple idea:

Great products are not built by design or engineering alone.

They are built when designers, developers, and product people understand each other better.

With two practical talks, Figma swag, food, and a packed room full of energy, Design Korea #0 showed that there is strong demand for English-friendly design and product conversations within Korea’s international tech community.

Building, learning, and connecting in Seoul

From the moment doors opened, the venue quickly filled up.

We kicked things off with check-in, food, and the usual relaxed Dev Korea atmosphere:

• designers meeting developers
• developers asking about design workflows
• people sharing what they are building
• new members joining the community for the first time
• familiar faces reconnecting

As always, the event was English-friendly and welcoming.

Don't worry about perfect English. Many attendees speak Korean too.
Come as you are. We're here to learn and connect together.

Figma MCP: When designers ship working prototypes, not frames

The first talk was by Inès Gruhier, founder of Odubu Design, a one-person design studio focused on UX/UI, design systems, branding, illustration, and product prototyping.

Her session focused on one of the most common sources of friction between designers and developers: the gap between what is designed in Figma and what actually gets implemented.

As Inès explained, the issue is not always a lack of skill on either side.

Designers and developers are often working with different tools, different mediums, and different assumptions. Figma shows what something should look like. Code shows how it should behave. And somewhere in that gap - between static mockups and actual interaction -miscommunication happens.

Inès showed how Figma MCP, combined with Claude Code, can help bridge that gap by letting AI read Figma files, understand the design system, color tokens, variables and layers, and use that to generate prototypes.

Instead of handing over static frames or long annotations, designers can create interactive prototypes that developers can inspect, test, and reference.

The session covered:

• Building an interactive debug panel to fine-tune a toggle animation
• Prototyping physics-based item drop animations for My Tiny Room that would be tedious to mock in Figma
• How she used Figma MCP to audit Bloxley's design system against the live codebase, surfacing mismatches and edge cases that existed on both the design and dev side
• Best practices for structuring Figma files so AI agents can interpret them correctly

One of the strongest takeaways was that this workflow does not replace designers or developers.

It changes the collaboration from interpretation to refinement.

Designers can finally show instead of tell, and developers can build from something real instead of guessing.

With AI handling the handoff, that shift lets both sides focus on what actually matters: making something that works, together.

From hypothesis to prototype: A solo designer's survival story

The second talk was by Anna Kim, a designer currently at Match Group, who shared a raw and honest story from her time as the solo designer at an early-stage AI startup.

Her talk was not about a perfect design process.

It was about survival.

Anna walked through how she helped rebuild an LLM app from scratch in 10 months, without a mentor, without an established design process, and with very little time.

At first, she expected to apply the textbook design process:

• user research
• personas
• journey maps
• structured brainstorming
• the classic double diamond

But startup reality looked very different.

There was too much ambiguity, too much raw feedback, and not enough time.

So instead of following the process exactly, Anna built her own.

She lived inside Reddit and Discord, tracked competitors, read user comments, DM'd users for quick interviews, and used AI to help categorize messy feedback into patterns her team could actually use.

Her key point was simple but powerful:

Intuition is not blind guessing.

It is the ability to make fast, informed decisions when you are deeply immersed in your users' reality.

That intuition led her to a major product decision: moving away from a simple chat interface and toward a node-based system that gave technical users more control over AI models, sessions, and workflows.

The session covered:

• how to survive as a junior solo designer in a fast-moving startup
• how to turn messy Discord and Reddit feedback into usable product insight
• why user intuition can become stronger when it is fed by raw data
• how prototypes can create alignment faster than long debates
• why designers sometimes need to skip parts of the textbook process and adapt to reality

Anna's talk was especially memorable because it was honest.

It showed the messy, uncomfortable, high-pressure side of design work that does not always appear in polished case studies.

Design and engineering belong in the same room

Compared to previous Dev Korea events, this one had a different energy.

The audience included many designers, but also plenty of developers, founders, and product-minded builders.

That mix made the conversations especially interesting.

Throughout the evening, people talked about:

• AI-assisted design workflows
• Figma MCP and code handoff
• how designers and developers can reduce miscommunication
• how to prototype faster
• how to use data without losing creativity
• how junior designers can grow in fast-moving teams

The level of engagement made one thing clear:

Design is not separate from tech.

It is part of how great products get built.

Figma swag, networking, and hallway conversations

Of course, the Figma swag also helped bring extra excitement to the evening.

But beyond the bottles, caps, tees, pins, and raffle prizes, the best part was still the people.

After the talks, attendees stayed to ask questions, meet new people, and continue the conversations that started during the sessions.

These informal conversations remain one of the most valuable parts of every Dev Korea event.

Huge thanks

This event would not have been possible without:

• Our speakers, Inès Gruhier and Anna Kim, for sharing practical, honest, and highly valuable design stories
• Figma for supporting the event and providing amazing swag
• BizCrush for supporting real-time English to Korean translation
• Maru by Asan Nanum Foundation for hosting us
• Cry Cheeseburger for providing delicious burgers for the attendees
• All the volunteers who helped make this event a success: Bertijn, Brian, Sonali and Suhyun but also everyone who made everything flow seamlessly from start to finish
• All 110+ attendees who filled the venue and brought great energy

Seeing the room packed for our first design-focused event was both surprising and incredibly motivating.

What's next

Design Korea #0 confirmed that there is strong demand for practical, community-driven design events in Seoul.

We'll keep bringing more events focused on real-world topics across engineering, design, product, startups, AI, and beyond.

Explore upcoming events: dev-korea.com/events

Want to stay connected?

• Follow us on LinkedIn: linkedin.com/company/dev-korea
• Follow us on X: x.com/dev_korea
• Watch previous talks: dev-korea.com/talks
• Join our Discord server: discord.gg/hqzMbuXy73
• Subscribe to the Weekly Dev Korea Digest Newsletter: dev-korea.com/newsletter

Here’s to more nights of learning, building, and meaningful community in Seoul. ✨

Ready for your next move?
Visit Dev Korea to explore the latest job openings at dev-korea.com/jobs, or if you're hiring, post a job at dev-korea.com/post-a-job and connect with our growing international tech community.

It also leaked 750 days of premium to 148 abusers.

This is the full story: what they exploited, what we missed, and the layered defense we shipped.

The referral system

Simple mechanic: share your code, friend signs up with it, both get rewarded. The reward tiers were:

Later we simplified to 1 referral = 1 day of full premium for both sides. No cap.

Detection (and what we got wrong)

We initially flagged 114 users with a referral_banned column on the users table. But that flag only blocked them from using referral codes — they could still sign in and use the app with all the premium time they'd already earned.

Worse: a banned user could delete their account and re-register with a different Gmail, getting a clean referral_banned = 0.

Their alt accounts — the fake Gmails they'd created to farm referrals — were completely untouched.

We'd banned the action, not the actor.

The fix: Layered defense

We shipped everything in one session — auth bans, deletion persistence, alt detection, premium revocation, and legal updates.

Layer 1: Auth-level ban

Added a single check at every authentication entry point:

export class AccountSuspendedError extends Error {
  constructor() {
    super("Account suspended for violating Terms of Service.");
    this.name = "AccountSuspendedError";
  }
}

// In AuthService
private assertNotBanned(user: User): void {
  if (user.referral_banned === 1) {
    throw new AccountSuspendedError();
  }
}

Placed before buildAuthResponse() in:

• googleSignIn() — blocks Google sign-in
• telegramLogin() — blocks Telegram bot sign-in
• telegramWidgetLogin() — blocks Telegram widget (web)
• refresh() — kicks already-logged-in users within 1 hour (token expiry)

The route handler returns 403 (not 401) with errorCode: "ACCOUNT_SUSPENDED":

function authErrorResponse(c: AppContext, e: unknown): any {
  if (e instanceof AccountSuspendedError) {
    return c.json(
      { message: e.message, errorCode: "ACCOUNT_SUSPENDED" },
      403
    );
  }
  return c.json({ message: getErrorMessage(e) }, 401);
}

Why 403 matters: our Android AuthInterceptor automatically retries on 401 (token refresh).
A 403 bypasses the retry loop and triggers immediate logout:

// AuthInterceptor.kt
if (response.code == 403 && !isPublicRoute) {
    Log.w("AuthInterceptor", "Account suspended (403). Logging out.")
    runBlocking { sessionManager.logout() }
    return response
}

Using 401 would have caused the app to silently retry in a loop, thinking it was an expired token. 403 tells the client: this isn't an auth problem, you're not welcome here.

Layer 2: Ban persistence across account deletion

When a user deletes their account, we snapshot their identity into a deleted_accounts table:

CREATE TABLE deleted_accounts (
    id TEXT PRIMARY KEY,
    googleId TEXT,
    email TEXT,
    telegramId TEXT,
    fcmTokens TEXT,  -- JSON array of device tokens
    referral_banned INTEGER NOT NULL DEFAULT 0,
    deletedAt TEXT NOT NULL DEFAULT (datetime('now'))
);

When anyone signs up, we check if their Google ID, email, Telegram ID, or FCM token matches a deleted account. If it does, we restore the ban using MAX():

UPDATE users SET
  referral_banned = MAX(referral_banned, ?)
WHERE id = ?;

MAX() is key — it means a ban can only be added, never accidentally removed. If the deleted account had referral_banned = 1 and the new account has referral_banned = 0, MAX(0, 1) = 1 — still banned. Delete and re-register with the same Google account? Still banned.

Layer 3: Alt Account Detection via Referral Graph

The 114 originally banned users were the fake accounts — the pawns. Their main accounts — the puppet masters collecting premium days — were still untouched. We walked the referral graph:

SELECT DISTINCT CASE
    WHEN u_referrer.referral_banned = 1 THEN r.referredUserId
    WHEN u_referred.referral_banned = 1 THEN r.referrerId
    END as alt_id
FROM referrals r
JOIN users u_referrer ON r.referrerId = u_referrer.id
JOIN users u_referred ON r.referredUserId = u_referred.id
WHERE (u_referrer.referral_banned = 1 AND u_referred.referral_banned = 0)
   OR (u_referred.referral_banned = 1 AND u_referrer.referral_banned = 0);

First pass: 30 alt accounts found.
Second pass: 4 more second-degree alts.
Third pass: 0 remaining. Clean.

The network wasn't deeply layered — most abusers were running simple one-hop schemes.
Final count: 148 banned users (114 + 30 + 4).

Layer 4: FCM device cross-check

Firebase Cloud Messaging tokens are unique per app install per device. When a user registers their device, we now check if that FCM token is associated with any banned user:

// In updateFcmToken()
const bannedDevice = await this.repo.findBannedUserByFcmToken(token, userId);
if (bannedDevice) {
  console.warn(
    `[Auth] Device shared with banned user: userId=${userId}, bannedUserId=${bannedDevice.id}`
  );
  await this.repo.banUser(userId);
}

SELECT u.id FROM devices d
JOIN users u ON d.userId = u.id
WHERE d.fcmToken = ? AND u.referral_banned = 1 AND u.id != ?
LIMIT 1;

This catches the scenario: banned user creates new Gmail, signs up on same phone, app sends FCM token → token matches banned user's device → new account auto-banned.

The known limitation: FCM tokens reset on app reinstall. A determined user can uninstall, reinstall, and get a fresh token. But we've raised the cost of abuse from "sign in with another Gmail" (30 seconds) to "uninstall, reinstall, create new Gmail, sign up fresh, and lose all progress" (minutes of effort for a few premium days). Most users are students studying for entrance exams — they're not going to keep grinding that loop.

Layer 5: Premium revocation

4 banned users still had active premium modules. All from referral rewards, not legitimate purchases (subscription_providerId: null). Revoked in one query:

UPDATE users SET
  active_modules = '[]',
  module_speaking_expiresAt = NULL,
  module_writing_expiresAt = NULL,
  module_reading_expiresAt = NULL,
  module_listening_expiresAt = NULL,
  subscription_tier = 'free',
  subscription_expiresAt = NULL
WHERE referral_banned = 1
  AND active_modules IS NOT NULL AND active_modules != '[]';

Layer 6: Legal coverage

Updated Privacy Policy and Terms of Service in all 3 languages (EN/RU/UZ):
Privacy Policy additions:

• Device identifiers used for abuse prevention
• Abuse flags retained after account deletion

Terms additions:

• Explicit prohibition on multi-account creation and referral manipulation
• Account suspension section with specific reasons
• Ban persistence clause: "Suspension status is retained even after account deletion and will be applied to any new account created from the same device or identity."

The data: Was the referral system worth It?

After the cleanup, we ran the numbers. The answer surprised us.

Genuine vs fraudulent

91% of referrals were real people referring real people. The system was working.

Retention: Referred vs organic

Referred users retained at 2.4x the baseline. Someone who joins because a friend told them is more committed than someone who found you on the Play Store.

Where did the premium days go?

749 premium days were given away. 693 (93%) went to fraudsters. Only 56 days went to legitimate referrers — roughly 8 weeks of one subscription's worth of premium given away in exchange for hundreds of sticky users. Most paid acquisition channels would kill for that ratio.

The fraud wasn't that the system was too generous. It's that the high-tier rewards attracted abusers who farmed at scale, while legitimate users mostly stayed at tier 1 (streak freezes).

Was the tiered structure the problem?

We briefly switched to a flat 1-day premium reward before shutting referrals down entirely:

The fraud rate barely changed. The problem was never the generosity of the reward — it was the tiered escalation. The reward-to-effort ratio had a cliff at tier 2 where 2 minutes of creating a Gmail suddenly bought you 7 days of premium. Under the 1-day system, you'd need a new fake account every day to maintain premium. The economics of farming collapsed.

The multi-account landscape

With referral abuse handled, we checked how many devices were running multiple accounts across the entire platform:

2,104 devices running multiple accounts — 4,971 accounts out of ~77K users (6.5%). The 5+ cluster (63 devices, 352 accounts) looked suspicious. But when we checked their trial flags and exam usage: all zeros. No trial abuse, no free exam farming. Every multi-account was built for referral farming, and with referrals dead, the accounts were inert.

We considered implementing ANDROID_ID fingerprinting and device-level account restrictions. Then we checked the data and decided not to build it. The problem was already solved. Ten minutes of SQL saved days of engineering.

Architecture summary

SIGN-IN ATTEMPT

• Google / Telegram / Widget
  → assertNotBanned()
  → 403 if banned
  → Token refresh also returns 403 (within 1 hour)

ACCOUNT DELETION

• Delete account → snapshot identity + ban flag
• Re-register with same identity
• If match found → restore ban via MAX()

DEVICE REGISTRATION

• Register FCM token
• If device linked to banned user → auto-ban

REFERRAL GRAPH

• Traverse referral chain
• Ban connected accounts (iterative)
• Revoke fraudulently earned premium

What we learned

1. Ban the actor, not just the action. Our initial referral_banned flag only blocked referral code usage. That's like banning someone from the loyalty program but still letting them shop. If someone is abusing your system, lock the door.
2. Deletion is an escape hatch. If you ban someone but let them delete and re-create, you haven't banned them. Snapshot identities before deletion and check on re-registration.
3. Walk the graph. The 114 banned accounts were the pawns. The 34 accounts collecting premium were the real abusers. Always check both sides of the referral relationship.
4. Tiered rewards create cliffs. Streak freezes at tier 1 attracted zero abuse. 7-day premium at tier 2 attracted all of it. The moment you give away the thing people pay for, someone will figure out how to farm it. Keep referral rewards non-fungible with your paid tier.
5. 403 vs 401 matters.** Using the wrong HTTP status code caused our Android interceptor to silently retry (thinking it was an expired token) instead of showing the user why they were blocked. Small detail, big behavioral difference.
6. Check the data before building defenses. We considered implementing ANDROID_ID fingerprinting, device-level exam quotas, and mandatory notification permissions. Then we checked: the 5+ account cluster had zero trial abuse, zero free exam farming. The only exploit was referral farming, which we'd already killed. Sometimes the best engineering decision is not to build something.
7. Don't kill a working growth channel because of a small abuse cluster. The fraud rate was only 9%. 91% of referrals were genuine users who retained at 2.4x baseline. 363 real users with 2x retention for 56 days of premium is an incredible acquisition deal. The system was working — it just needed guardrails, not demolition.

Connect with Azizbek

• LinkedIn: /in/typos-bro

Built by MilliyTechnology. We're building AI-powered exam prep tools for students in Central Asia. If you're working on similar problems, reach out.

If you want to be next and contribute, send us an email at florian@dev-korea.com.

Ready for your next move?
Visit Dev Korea to explore the latest job openings at https://dev-korea.com/jobs, or if you’re hiring, post a job at https://dev-korea.com/post-a-job and connect with our growing international tech community.

With a highly engaging talk on AI and serverless, 46 pizzas, and a room full of energy, this event once again showed how strong the demand is for practical, real-world tech conversations within Korea’s international developer community.

Building, learning, and connecting in Seoul

From the moment doors opened, the venue quickly filled up.

We kicked things off with check-in and food, and with 46 pizzas provided by Dev Korea, conversations started immediately:

• people sharing what they are building
• discussing AI trends and tooling
• reconnecting with familiar faces
• meeting new members of the community

As always, the atmosphere was relaxed, friendly, and welcoming.

Don’t worry about perfect English. Many attendees speak Korean too.
Come as you are—we’re here to learn and connect together. 🇰🇷

🧠 How to efficiently develop and deploy your AI-based product with serverless

The evening featured a single, deep-dive session focused on one of the most relevant topics today: AI infrastructure and serverless architecture.

Sang (Chris) Park delivered an awesome, highly engaging talk exploring how to build and ship AI-powered products efficiently using serverless.

The session covered:

• why infrastructure matters just as much as the application layer
• what serverless architecture actually is and when to use it
• how serverless helps remove bottlenecks for fast-moving teams
• why it is especially well-suited for AI-based products
• a live, real-time demonstration of building and deploying

The talk struck a strong balance between concept and practice, making it valuable for both experienced engineers and builders exploring AI for the first time.

🌱 AI and serverless: a growing focus in the community

Compared to previous events, this meetup clearly reflected the current shift in the ecosystem.

Conversations throughout the evening focused on:

• building AI-powered products in practice
• choosing the right infrastructure for fast iteration
• balancing cost, scalability, and developer experience
• how serverless changes the way teams ship

The level of engagement, and the size of the crowd, made one thing clear:
AI and modern infrastructure are top of mind for the Dev Korea community.

🍕 Pizza, networking, and hallway conversations

Food once again played a key role in setting the tone.

With 46 pizzas shared across 160+ attendees, the event felt lively from start to finish.

Throughout the night, people:

• exchanged ideas around AI tools and workflows
• discussed serverless setups and real-world use cases
• talked about startups, side projects, and careers
• built new connections and strengthened existing ones

These informal conversations remain one of the most valuable parts of every Dev Korea event.

🙌 Huge thanks

This event would not have been possible without:

• Our speaker Sang (Chris) Park for an awesome, super engaging talk on AI and serverless
• The AWS Korea team, Yongho Choi, SeongHan Kim, Sewoong Kim, for helping the event run smoothly even with such a large crowd
• Our amazing volunteers: Aleksey Kim, Bertijn Pauwels, Ian Wagner, Kathie Lee, Loïc Castelin, Myungjun Jung, Rob Sawyer, and many more who helped behind the scenes
• All 160+ attendees who filled the venue and brought incredible energy

Seeing the space completely packed was both surprising and incredibly motivating for us as organizers.

👀 What’s next

Dev Korea continues to grow, and Dev Korea #7 confirmed the strong demand for practical, community-driven tech events in Seoul.

We’ll keep bringing more events focused on real-world topics, from AI and infrastructure to product, startups, design, and beyond.

👉 Explore upcoming events: https://dev-korea.com/events

Want to stay connected?

• Follow us on LinkedIn: https://www.linkedin.com/company/dev-korea
• Follow us on X: https://x.com/dev_korea
• Watch previous talks: https://dev-korea.com/talks
• Join our Discord server: https://discord.gg/hqzMbuXy73
• Subscribe to the Weekly Dev Korea Digest Newsletter: https://dev-korea.com/newsletter

Here’s to more nights of learning, building, and meaningful community in Seoul. 🍻✨

Ready for your next move?
Visit Dev Korea to explore the latest job openings at https://dev-korea.com/jobs, or if you’re hiring, post a job at https://dev-korea.com/post-a-job and connect with our growing international tech community.

This article was originally published by Maëlle Dairain on her LinkedIn and is reposted here with her permission. Thanks to Maëlle for letting us share it with the community.

Most foreign professionals in Korea assume the long-term path is linear:

D-2 → D-10 → E-7 → F-2-7 → F-5
That's the common narrative.

In February 2026, I moved directly from D-10 to F-2-7, without ever holding an E-7 visa. I'm sharing this not as a "hack," but as a case study in preparation, timing, and understanding how the system actually works.

🌏 My background

Hi I'm Maëlle, a 24 year-old French woman who discovered South Korea for the first time during an exchange semester in Sungkyukwan University 3 years ago. After graduating from a Master's degree in Engineering (Smart and Resilient Cities) and while working a year in urban project development at Linkcity (Bouygues Construction Group) in France, I made the deliberate decision to come back.

I returned to Korea in April 2024 on a Working Holiday (H-1) visa with a clear goal: build a long-term career here. During that year, I focused on:

• Improving my Korean (until TOPIK Level 4)
• Building a professional network
• Securing relevant work experience
  In April 2025, I transitioned to a D-10-1 (job-seeking visa). After a short-term consulting experience at GGGI, I joined Yooshin Engineering Corporation as an intern in September 2025 and signed a one-year contract in January 2026.

One important detail: switching from H-1 to D-10 cannot be done in-country — Korea does not allow this change of status domestically. I ended my H-1 a little earlier than planned, spent two days in Japan, and re-entered Korea as a tourist to apply for the D-10 upon arrival.

First: Understand which pathway applies to you

Before anything else, I want to be clear: this article describes one specific pathway to the F-2-7 visa. The system has multiple routes, and most people will not qualify through the same one I used.

The three main (but not only) eligibility pathways are:

1. The standard route (most common): You hold a professional visa such as E-7, E-2, E-3, or similar, have been legally residing in Korea for at least 3 consecutive years under that status, and accumulated enough points. This is what most foreign professionals eventually use. It takes time, but it's well-documented and reliable.
2. Korean domestic graduate: You completed a Master's degree or higher at a Korean university, have been residing in Korea for at least 3 consecutive years (with D-2 or D-10-1 visa), and have confirmed employment or are employed under a qualifying professional visa. This is a faster route for those who studied here long-term.
3. KOSPI/KOSDAQ employee: You are employed or have confirmed employment at a corporation listed on the KOSPI or KOSDAQ stock markets, in a managerial or professional occupation per the Korean Standard Occupational Classification. If you meet this condition and have enough points, the usual 3-year residency requirement is waived entirely.

🎯 Why I chose F-2-7 over E-7

Honestly, E-7 was my original plan. It's the default answer most people give when you ask "what visa do I get once I have a job in Korea?"

But while researching the E-7 → F-2 pathway, I came across a specific provision I hadn't heard of before: foreign professionals who are employed at a KOSPI or KOSDAQ-listed company may be eligible for the F-2-7 visa directly.

I checked: Yooshin is listed on KOSDAQ!

That changed everything. F-2-7 offered something E-7 couldn't in the short term:

• Independence from employer sponsorship
• Freedom to change jobs without losing status
• A direct path toward long-term residency
• For someone building a career here on their own terms, it was clearly the better option.

✅ What made it possible

The F-2-7 is a points-based system. Here's a rough breakdown of where my points came from:

• Education: Master's degree in engineering
• Age: Age bracket gives more or less points
• Korean language: TOPIK Level 4
• Income: Annual salary (1-year contract)
• Special condition: Employed at a KOSDAQ-listed company

Note: under the standard pathway, income points are calculated from your most recent 소득금액증명서. But under the KOSDAQ/KOSPI pathway, you should not have one, so income is assessed directly from your employment contract instead.

The KOSDAQ condition is significant because it essentially fast-tracks eligibility for professionals at listed companies — but it comes with strict conditions, one of which nearly derailed my application entirely...

⚠️ The obstacle I didn't see coming

I had spent months preparing every document and I went to the immigration office confident.

The officer asked whether I had any declared income in the past two years (here 2023 and 2024). The KOSDAQ-employee pathway requires that you cannot submit a 소득금액증명서 (income certificate) showing any earnings in that period. At the time of my application, the 2025 income certificate was not yet available — it only becomes accessible from May of the following year. That meant immigration needed to verify the two previous fiscal years: 2023 and 2024.

I assumed I didn't have one — I had never held a formal work visa in Korea.

They sent me to the tax office to verify. And there it was: a 소득금액증명서 from 2024 showing an amount under 70,000 KRW. Where did it come from? During my H-1 year, I had animated two cooking classes (two hours each, just giving instructions) for which I was paid a small fee. That was it!

The officer was clear: it doesn't matter how low the amount is. Any recorded income (a part-time shift, a paid SNS post, a small gig) will appear on your 소득금액증명서 and can disqualify you under this specific condition. The amount is irrelevant; the existence of the entry is what counts.

After months of preparation, I walked out of that office without a visa…

The follow-up that changed the outcome

A week later, I went back — this time intending to apply for an E-7.

But my intuition told me to bring up the income certificate issue again and explain the full context. I spoke with a different officer. She reviewed my documents, consulted with a colleague, and acknowledged that while the previous interpretation was technically correct, the amount was so negligible that they were willing to exercise discretion and accept the application.

I submitted for F-2-7. It was approved one week later and my new 외국인등록증 (residence card) arrived within two weeks.

Note: because my situation was non-standard, I was granted a 1-year visa rather than the 2-year validity I would have been eligible for based on my salary.

💡 Practical advice for D-10 holders

If you're on a D-10 and trying to figure out your next step, here's what I'd suggest:

• Don't assume E-7 is your only option. If your employer is listed on KOSPI or KOSDAQ, verify whether you qualify for the alternative pathway.
• Run your point simulation early. Understand which documents will be scrutinized most
• Check your tax record before applying. Even minor paid activity during H-1 or D-10 (a part-time job, a cooking class, a paid SNS post, anything) can appear on your income certificate. The amount is irrelevant — the record is what matters.
• Prepare documentation meticulously. Translations, apostilles, contract clarity — details determine outcomes.
• Be persistent, but also be strategic. Avoid being emotional. How you present your case matters.

Risks & honest limitations

This process is not straightforward, and I want to be clear about a few things:

• Immigration officer discretion is real. What one officer interprets strictly, another may handle differently. This creates uncertainty that no amount of preparation can fully eliminate.
• Documentation requirements are strict and specific. Translation standards, apostille requirements, and document formats are not flexible. One missing element can delay or block your application.
• Edge cases like mine may not resolve in your favor. The outcome I described depended on a specific officer being willing to apply judgment. That is not guaranteed.

The F-2-7 via KOSDAQ/KOSPI provision has conditions that need to align precisely. If your income record, contract terms, or point calculation fall slightly outside the criteria, you may need a different pathway.

Final thought

The Korean immigration system is structured, but it's not inflexible. There are pathways that aren't widely advertised, and eligibility conditions that reward people who do their research carefully.

My path from D-10 to F-2-7 wasn't linear. It involved a setback, a second attempt, and a conversation that could have gone either way. What made the difference was preparation, knowing the system well enough to ask the right questions, and the willingness to try again.

If you're planning your next step from D-10 or E-7, research the pathways early — small details can change the entire timeline.

Connect with Maëlle

• LinkedIn: /in/maelledairain
• Linktree: /maelledairain

If you want to be next and contribute, send us an email at florian@dev-korea.com.

Ready to explore your next move? Visit Dev Korea to check the latest job openings, or if you're an employer, post a job and connect with our vibrant community of tech talent eager to contribute to Korea's innovation ecosystem.

Profile snapshot

• Name: Matheus Miranda
• Nationality: Brazilian/British
• Current role & company: AI Engineer at Tova Earth
• Years in Korea: 12
• Visa status: F-2 (D-2 -> E7 -> F-2)
• Languages spoken: English, Portuguese, Spanish, Basic Korean
• Previous countries worked in (if any): England

Background & Career

Q. Can you briefly introduce yourself and your current role?

Matheus: My name is Matheus. I'm 36 years old and originally from Brazil.

My story in Korea began in 2013, when I was awarded a scholarship to study my Master's degree in Electrical and Computer Engineering at Seoul National University. Since then, I've had over 10 years of work experience as an engineer in Korea. I'm currently a Senior AI Engineer at a startup called Tova Earth.

Q. What was your career path before moving to Korea?

Interestingly, my entire professional career has been in Korea. Before moving here, the only job I had was as a salesperson at a phone shop during my gap year — not exactly related to my current career path.

Q. What inspired you to work in Korea?

The scholarship was offered by Samsung Electronics, and as part of the package, they included a two-year work contract in a department related to my research topic. That opportunity was extremely attractive to me, as it allowed me to gain early industry experience at one of the most well-known consumer electronics and mobile phone companies in the world.

The Samsung Global Scholarship was a fantastic opportunity overall, and it was the main reason I was excited to come to Korea.

Experience Working in Korea

Q. Tell us briefly about your company (size, industry focus, notable achievements).

Tova is a relatively new startup headquartered between NYC and London (I work remotely from Seoul). It's a small team of eight people, including myself. The company focuses on leveraging deep learning and other AI technologies for water risk management.

Our main customer segments include water utility companies and private-sector companies in food and beverage, which require large volumes of water for operations. Tova has participated in the Techstars London accelerator, raised angel funding, and has paying customers, including a water utility company in California.

My role involves designing and deploying AI systems.

Q. How did you find your current job? Was the hiring process challenging as a foreigner?

I generally use LinkedIn, Dev Korea, and other remote job boards to discover open positions. In this case, however, I was contacted directly by the CEO of Tova. We previously worked together on a summer volunteering project in Peru during our undergraduate studies in London.

The hiring process was fairly informal and based largely on trust and past accomplishments. Since the company is international and remote-first, there were no specific challenges related to being based in Korea.

Q. Can you describe your typical workday?

My day involves coding with Claude Code, evaluating and monitoring agent performance, and deploying systems on AWS.

I typically take a break in the afternoon to go to the gym and have dinner, then work again in the evening so I can overlap with team members in the UK and US time zones.

Q. How does Korean work culture compare to your home country's?

Since I have limited experience working in Brazil or the UK, I may not be the best person to make a full comparison. However, I did get a glimpse of the UK work culture while participating in the Techstars London accelerator with one of my startups (yes, I was a co-founder at two startups).

In my experience, UK work culture encourages creativity and open expression of ideas, regardless of title or seniority. Professional criticism is often viewed positively and contributes to constructive discussion.

In Korea, hierarchical culture is strongly present in many companies. There is often a strong emphasis on following instructions and working long hours. Of course, this doesn't apply to every Korean company, but it is a pattern I have observed.

I was surprised by the speed at which things move in the Korean tech sector, especially in smaller companies. It’s a very dynamic work environment that keeps adapting to new trends.

Q. Do you speak Korean at work? How is communication handled in your team?

I have never spoken Korean at work. Even during my time at Samsung, I was encouraged to use English to help colleagues practice and to support Samsung's vision of becoming a global company.

Q. What challenges have you faced adapting to Korea's work environment?

The communication hierarchy can be frustrating, as it sometimes slows productivity. I once had to start a long email chain to request information from someone sitting just two desks away, because direct communication wasn't considered appropriate.

I have also experienced a strong sense of 눈치 (nunchi), which sometimes prevented me from expressing my opinions openly. Finally, the "ppalli-ppalli" culture can lead to rushed decision-making without sufficient discussion or alignment.

Lifestyle & Community

Q. What's your life outside of work like? How have you built your social life here?

I've been fortunate to build strong friendships since arriving in Korea in 2013. Although people come and go, I managed to keep a good friendship with people I met during my time at Samsung and every company since, as well as other "Korea OGs" that I met along the way.

Outside of work, I enjoy outdoor activities with my fiancée, my dog, and friends. We also like hosting dinners at home and organizing group outings. Personally, I enjoy cooking and photography.

Q. How much has knowing Korean (or learning it) helped in your daily life?

If my Korean were better, it would help tremendously. Being able to have deeper conversations with neighbors and colleagues significantly improves quality of life and helps you feel more integrated into the community.

Q. What was your biggest culture shock when you first arrived?

One shock was how readily people said "yes" to their superiors.

Another memorable moment happened during my time as a student at SNU. My labmates expected everyone to remain in the research lab from 9 a.m. to 5 p.m., unless attending class. They told me that as long as I followed this norm, they would support me whenever needed.

One afternoon, while preparing for an exam, I decided to study at the library instead. The next day, senior PhD students pulled me aside to express their disappointment that I had left the lab without notifying them. That experience taught me a lot about expectations within group settings.

Reflections & Advice

Q. Do you plan to stay in Korea long-term, and how do you see your career evolving here?

I can certainly see my career continuing to evolve in Korea. I’ve recently started taking Korean language classes again so that I’ll be better prepared in case I decide to work for a Korean company in the future.

At the moment, I’m very happy working remotely while living in Seoul, and I can see myself continuing this lifestyle for quite some time.

Q. What tips would you give to someone searching for a tech job in Korea?

Pay attention to hiring cycles and prepare thoroughly for technical tests. Research each company carefully to understand what types of questions they might ask, and prepare strong answers for the specific role.

Q. What can Korean companies do to better support and integrate international employees?

Reduce communication barriers related to rank and age. Provide clear onboarding about Korean culture for international employees, keeping in mind that things that feel natural to Koreans can be confusing or unintuitive for outsiders.

Q. What's one thing you wish you knew before moving here?

The importance of deeply understanding the culture beforehand, so that I could have been better prepared to navigate the work environment and integrate into society more smoothly.

Q. Would you recommend Korea to other international tech professionals, and why?

Yes, in Korea you will meet highly talented colleagues, and be involved with cutting-edge technologies. For those open to adapting to cultural differences, it can be a very rewarding experience both professionally and personally.

Rapid Fire

• Favorite Korean food: Charcoal-grilled dakgalbi (숯불 닭갈비)
• One must-visit place in Korea: Han River (한강)
• Most surprising thing about Korean tech culture: People are very intelligent and kind
• Favorite Korean word or phrase: 화이팅 (Hwaiting meaning "Good luck," "You can do it," or "Fighting!")

Connect with Matheus

• LinkedIn: /in/matheus-f-miranda

If you want to be next and contribute, send us an email at florian@dev-korea.com.

Ready to explore your next move? Visit Dev Korea to check the latest job openings, or if you're an employer, post a job and connect with our vibrant community of tech talent eager to contribute to Korea's innovation ecosystem.

From the moment doors opened, attendees gathered around food, reconnected with familiar faces, and met new people from across the Kubernetes and platform engineering ecosystem. As always, the mix of backgrounds, local and international, junior and senior, individual contributors and founders, made the discussions richer and more grounded.

One reminder we always emphasize, and that clearly resonated again this time:

Don’t worry about perfect English. Many attendees speak Korean too.
Come as you are. We’re here to learn and connect together. 🇰🇷

🧠 How to survive and thrive in a multi-cluster world

The evening opened with a deep dive into one of the most relevant challenges modern teams face.

Ivan Porta explored how organizations are moving away from large, monolithic Kubernetes clusters toward multi-cluster environments, using entire clusters to isolate teams, workloads, or environments.

Rather than treating multi-cluster as a buzzword, the talk focused on practical reality:

• why teams adopt multi-cluster setups in the first place
• how service meshes and GitOps workflows help manage complexity
• the reliability and operational gains that are possible
• trade-offs between multi-cluster and multi-cloud approaches
• how emerging ideas like federated Services fit into the picture

The session resonated strongly with engineers already feeling the limits of single-cluster thinking, and offered a clear mental model for navigating the multi-cluster world without losing control.

🎥 Watch the talk

🚀 Tilt for local Kubernetes microservices development

Next, the focus shifted from production architecture to the daily reality of development.

Anthony Corbacho tackled a familiar pain point: local Kubernetes development that slowly drifts away from production, leading to "it works on my machine" surprises.

The talk demonstrated how Tilt helps teams:

• run real microservice stacks locally on Kubernetes
• keep the same Kubernetes resources and configuration model as production
• rebuild and redeploy only what changed
• centralize logs, status, and service dependencies in one place

For many attendees, this session connected the dots between developer experience and platform consistency, showing how tooling choices directly affect team velocity and confidence.

🎥 Watch the talk

🌱 A first infra-focused Dev Korea event

Dev Korea #6 marked an important milestone for us: our very first event fully focused on infrastructure, Kubernetes, and containers.

Until now, most Dev Korea meetups leaned more toward product, startups, and application-level topics. This time, the conversation shifted clearly toward:

• infrastructure and platform engineering challenges
• running Kubernetes in real production environments
• developer experience and tooling for complex systems
• balancing reliability, scalability, and team velocity

For many attendees, it was refreshing to dive deep into infra topics in a community setting that still felt approachable and welcoming. The level of engagement, both during the talks and in the hallway conversations, showed there is strong interest in more infrastructure and cloud focused events within the Dev Korea community.

This first infra-focused meetup set a solid foundation, and it's very likely not the last time we'll explore Kubernetes, containers, and platform engineering together.

🍔 Food, networking, and hallway conversations

Food was served during check-in, setting a relaxed tone from the start and making it easy for conversations to begin naturally.

Throughout the night, people:

• exchanged experiences running Kubernetes in production
• compared local development setups
• talked about startups, side projects, and career paths
• made new connections and strengthened existing ones

These informal discussions are always a core part of Dev Korea and often where the most valuable learning happens.

🙌 Huge thanks

This event would not have been possible without:

• Buoyant and Cry Cheeseburger for sponsoring this event
• Our speakers Ivan Porta and Anthony Corbacho for sharing practical, experience-driven insights
• All the volunteers who helped make this event a success: Aaron McCollum, Bertijn Pauwels and Rob Sawyer who made everything flow seamlessly from start to finish
• All the attendees who brought thoughtful questions and great energy
• Everyone who helped make the evening welcoming, smooth, and enjoyable

Seeing engaged conversations continue well after the talks ended is always the best signal that an event worked.

👀 What’s next

Dev Korea continues to grow, and Dev Korea #6 reinforced the demand for practical, technical, community-driven meetups in Seoul.

We’re already preparing upcoming events with global tech companies and local builders—covering real tools, real systems, and real lessons learned.

👉 Explore upcoming events at https://dev-korea.com/events
👉 Watch or rewatch the talks at https://dev-korea.com/talks

Want to stay connected?

• Follow us on LinkedIn: https://www.linkedin.com/company/dev-korea
• Follow us on X: https://x.com/dev_korea
• Join our Discord server: https://discord.gg/hqzMbuXy73
• Subscribe to the Weekly Dev Korea Digest Newsletter: https://dev-korea.com/newsletter

Here’s to more nights of learning, building, and meaningful community in Seoul. 🍻✨

Ready for your next move? Visit Dev Korea to explore the latest job openings at https://dev-korea.com/jobs, or if you’re hiring, post a job at https://dev-korea.com/post-a-job and connect with our growing international tech community.

"빨리 빨리!" (Ppalli Ppalli!)

It literally means "quickly, quickly."
But culturally, it means much more.

1. What is "Ppalli Ppalli"?

"Ppalli" means fast in Korean. Saying it twice emphasizes urgency.
Over time, it became shorthand for a cultural mindset: do it fast, move quickly, don't waste time.

Speed is especially valued in the service industry. Food arrives quickly. Deliveries come the same day. Public services move efficiently.

Fun fact: Korea's country code is +82.
In Korean pronunciation, 8 is "pal" and 2 is "i", which sounds a bit like "pal-i" (ppalli).
It's not the origin of the term, but many Koreans enjoy the coincidence.

2. Where did it come from?

Some Korean linguists suggest that "ppalli ppalli" culture may be linked to Korea's strong tendency toward uncertainty avoidance — the desire to reduce ambiguity and instability as quickly as possible.

Historically, the Korean peninsula experienced frequent invasions and instability. Later, during the rapid industrialization of the 1960s and 70s, speed became a national priority. Light industry and heavy chemical industries grew rapidly, and both the government and citizens embraced efficiency and acceleration.

Speed wasn't just a preference.
It became a survival strategy, and eventually, a competitive advantage.

3. Everyday "Ppalli Ppalli" moments

What Serin notices

• Pressing the elevator "close" button repeatedly until the door fully shuts
• Honking if the car in front doesn't move within one second after the light turns green
• Standing up and opening overhead bins before the airplane has fully landed

What Florian notices

• The bus driver departing immediately the moment the doors close
• Delivery drivers sometimes bending traffic rules to deliver faster
• Switching supermarket lines constantly to find the one that moves fastest

Shared on our Discord community

• Customers shouting their order before even sitting down
• Passports or visas being issued surprisingly fast

These are small things, but together, they paint a picture.

4. When foreigners start saying "ppalli" too

Interestingly, this mindset does not stay foreign for long.

Many foreigners who live in Korea for years gradually adapt to the "ppalli ppalli" rhythm, often without even realizing it. What initially feels rushed or stressful slowly becomes normal. Waiting starts to feel longer. Delays feel heavier. Speed becomes the default expectation.

The real surprise comes when they return to their home country.

Suddenly, everyday life feels slow.
Customer service takes too long.
Processes feel inefficient.
People seem oddly relaxed about delays that now feel unnecessary.

What once felt exhausting in Korea now feels efficient.
What once felt calm back home now feels frustrating.

In that sense, "ppalli ppalli" is not just a cultural trait you observe. It is a tempo you internalize. Living in Korea quietly recalibrates your sense of time.

Many long-term foreigners end up carrying Korea's speed with them, even after they leave.

5. The advantages of "Ppalli Ppalli"

There's a reason many foreigners are impressed by Korea's speed.

• 🚇 Public transportation runs precisely and reliably
• 🍜 Food and delivery services are incredibly fast
• 🏢 Administrative processes (banking, visas, paperwork) can be highly efficient

In industries with short life cycles, like food service or consumer tech, speed itself becomes part of the value proposition.

6. The downsides

However, speed can also come at a cost.

• People who work quickly, even if slightly inaccurate, may be perceived as more competent
• This mindset can contribute to safety negligence or rushed construction. Several major accidents in Korea, including the Seongsu Bridge collapse or the Sampoong Department Store disaster, are often cited as reminders of what can happen when speed outweighs safety.
• Those who cannot move quickly (children, elderly, or simply people with different working styles) may feel left behind
• Mistakes or failures may be seen as "wasted time" rather than learning opportunities

When speed becomes the primary measure of performance, quality and reflection may suffer.

7. Conclusion: Speed vs. quality in the AI era

Korea's "ppalli ppalli" culture has undeniably fueled economic growth and industrial success.

In fast-moving service industries, speed can be attractive and efficient.
But in industries where craftsmanship, safety, or long-term reliability matter, an excessive focus on speed may reduce overall quality.

In today's IT world, especially with AI accelerating software development cycles, the temptation to move faster than ever is strong.
Yet differentiation increasingly comes not just from speed, but from quality, depth, and thoughtful execution.

Perhaps the future isn't about choosing between speed and quality.

It's about knowing when to move fast, and when to slow down.

Ready to explore your next move? Visit Dev Korea to check the latest job openings, or if you’re an employer, post a job and connect with our vibrant community of tech talent eager to contribute to Korea’s innovation ecosystem.

Profile snapshot

• Name: Tabassum N Haque
• Nationality: Bangladeshi
• Current role & company: Managing Director at Katriyam
• Years in Korea: 13+ years
• Visa status: F5 (Permanent Resident)
• Languages spoken: English, Korean, Bengali
• Previous countries worked in (if any): N/A

Background & career

Q. Can you briefly introduce yourself and your current role?

Tabassum: I came to Korea back in 2012 as a Korean Government Scholarship Grantee to pursue my master's degree in Computer Engineering. I have worked in the field of Blockchain and Web3 since 2015, mostly focusing on strategic planning and business infrastructure for the blockchain ecosystem.

Currently I have my own web agency called Katriyam where we provide services to mostly Korean clients. I also work as a blockchain consultant and institutional business developer for BaaS (Blockchain as a Service).

Q. What was your career path before moving to Korea?

Initially I wanted to be in Academia. So, I was mostly working in research labs experimenting with new ideas.

Q. What inspired you to work in Korea?

Honestly, it was mainly because I took it as a challenge. When I was doing my masters, I wanted to go for a PhD after that and eventually become a professor. However, I kept hearing that I chose the path everyone else from Bangladesh in Korea chooses and that I should stick to it, because the corporate world in Korea is very hard and I have to struggle a lot. That initially gave me motivation to actually find a job in Korea.

But honestly, I have to say that I was lucky that I landed an internship in a company that was trying to implement Bitcoin in their e-commerce platform back in 2015. They were way ahead of their time, but it opened a whole new world to me.

Experience working in Korea

Q. Tell us briefly about your company (size, industry focus, notable achievements).

Katriyam is still a small startup. We are a team of 7 people and we focus mostly on Web/App design and development. We also provide A-Z blockchain project incubation starting from project ideation, preparing whitepapers, investor pitch deck, investor connection, consultation about the blockchain ecosystem and which main-net will be suitable for a certain project, writing smart contracts, token listing and beyond. Apart from that, we also have in-house AI experts who have worked and delivered a few interesting AI projects.

Q. How did you find your current job? Was the hiring process challenging as a foreigner?

I can talk about the recent contractual job as a Blockchain Business Developer which I just finished. I think my experience will be a little different than other foreigners. My field is fairly new and most of the companies who are in blockchain and Web3 are startups or SMEs who are expanding a subsidiary to the Web3 field.

As someone who has 10 years of experience in a niche field and can speak both Korean and English, the hiring process doesn't get that hard for me. I mostly never get rejected if I get a job interview. However, getting an interview is the hardest challenge. There are a few job portals where most Korean companies post their jobs, however, they don't mention whether they want a Korean or a foreigner. So, even though I might have the said qualification, I can still get rejected because of not being Korean.

Q. Can you describe your typical workday?

Since I was working at an on-site job until recently, I can share the typical work day based on that. I usually wake up, go to the gym, walk my dog and then get ready for work. I live in Suwon and my workplace is in Seoul. So I have a long commute of 3 hours a day. After I go to work I usually check emails to see if I’ve received any emails from potential partners. I go on to have a few meetings a day both in-house and with clients. As my field's market trend changes on a weekly basis, market research is something which is very important for me. Based on research, we might have to take some decisions to make some tweaks in our market reach. After the whole day of work, many times I have networking events to attend for my own company Katriyam. If I don't have networking events, I get back home and talk to my co-founder, check on the work status of my company and then head to bed.

Q. How does Korean work culture compare to your home country's?

Since I am from Bangladesh which is also in Asia, the work culture isn't that different honestly. We also have a strong hierarchy based work culture. However, the biggest difference is the overtime working culture in Korea. In Bangladesh, people can leave as the clock hits 5pm. But in Korea, staying for longer because the seniors haven't left is something very interesting. Koreans even have a word for it called "Nunchi".

Q. Do you speak Korean at work? How is communication handled in your team?

Yes I mostly use Korean at work. Most of the companies I have worked for are mostly functional in Korean. However, when I am handling overseas partners, I use English.

Q. What challenges have you faced adapting to Korea's work environment?

One of the challenges which most foreigners face is to blend in with Koreans as one of them. They have a hard time accepting that we foreigners are working alongside them although they also need us as much as we need them. So, in the initial days, I tried to mould myself according to their style both mentally and physically. I stopped looking very "Bangladeshi" as it was hard for them to understand the cultural differences. However, as time passed and I made a place for myself, I eventually started being who I am and I hardly face any judgment anymore.

Lifestyle & community

Q. What's your life outside of work like? How have you built your social life here?

Yes, my social life is pretty strong in Korea. I have a very healthy social circle as well as, I meet new people very often to expand my network. I usually have 2~3 social meetups every week. Apart from that, I also like to stay active in terms of working out, volunteering and giving back to Korea which has given me a lot.

Q. How much has knowing Korean (or learning it) helped in your daily life?

Knowing Korean has honestly been life-changing. Not only did it open a lot of new doors of opportunities, it also helped me to understand Koreans to their core. You cannot expect to know a country's people without talking to the people of all levels of a country. I have learnt more about Korea and the emotions of people from randomly talking to street vendors, restaurant owners, taxi drivers and so on. They give you the perspective of a country more than a corporate official.

Q. What was your biggest culture shock when you first arrived?

There were many actually. But if I have to highlight one, it has to be the "ppalli ppalli" culture. As someone coming from a very laid back society, it was overwhelming initially. However, I think I am one of those "ppalli ppalli" people now.

Reflections & advice

Q. Do you plan to stay in Korea long-term, and how do you see your career evolving here?

I definitely see having a tie with Korea all my life. I cannot really say I will be here forever or not, but as I have established businesses which are directly tied to Korea, I definitely think I will always be working with Korea and Koreans whether I stay here or leave.

Q. What tips would you give to someone searching for a tech job in Korea?

Making a strong portfolio is very crucial if you are searching for a tech job in Korea. Also, it can be unpopular advice, but take the first low-paid job you are offered. Not because of the money, just to understand Korean work culture so that you know how to move forward with your plan.

Q. What can Korean companies do to better support and integrate international employees?

I honestly think Korean companies need to be aware of the visa process of hiring a foreigner in Korea. A lot of times they promise a visa and then they cannot because they thought it would be easy to provide a visa.

Second would be understanding that different foreigners are coming from different countries. They cannot expect all of us to act a certain way because we all are "foreigners". Being open to understanding the cultural difference and being open to accepting it is very crucial.

Q. Would you recommend Korea to other international tech professionals, and why?

Definitely! Korea can be very hard in terms of work environment and strong hierarchy. But we cannot deny the fact that working in the Korean tech industry is like working for the future! The future which has a higher chance to come into reality and not just "fake promises". The innovations happening in Korea are something the world is looking for. So, if you are tech-savvy, it is always great to try diving into the corporate and business world of Korea.

Rapid fire

• Favorite Korean food: Dak-galbi
• One must-visit place in Korea: Chuncheon (underrated)
• Most surprising thing about Korean tech culture: That one time when the bank was actually providing a loan to buy land in the Metaverse!
• Favorite Korean word or phrase: ppalli ppalli

Connect with Tabassum

• LinkedIn: /in/tabassum-n-haque
• Linktree: /tabassum90

If you want to be next and contribute, send us an email at florian@dev-korea.com.

Ready to explore your next move? Visit Dev Korea to check the latest job openings, or if you're an employer, post a job and connect with our vibrant community of tech talent eager to contribute to Korea's innovation ecosystem.

Profile snapshot

• Name: Sahar Ayachi
• Nationality: Tunisian
• Current role & company: Chief Strategy Officer (CSO) at VanaM Inc.
• Years in Korea: ~ 7 years (3 as student, almost 4 as employee)
• Visa status: E
• Languages spoken: Arabic, French, English, Korean, Spanish
• Previous countries worked in (if any): Germany

Background & career

Q. Can you briefly introduce yourself and your current role?

Sahar: My name is Sahar, I'm from Tunisia and I am currently the CSO at VanaM Inc., a South Korean startup specialized in thin film deposition where I lead global strategy and business development. My role covers international partnerships, investor relations and the establishment of VanaM's first international unit.

Q. What was your career path before moving to Korea?

I originally came to Korea as a transfer student midway through my bachelor's degree in engineering and completed my three years here. After that, I pursued a master's degree in Japan and later a PhD in Germany. I returned to Korea as a Research Professor with the intention of continuing in academia, until my career goals evolved.

Q. What inspired you to work in Korea?

Having lived in Korea before as a student and having experienced life and work in other countries, I felt that my chapter with Korea wasn't finished. That sense stayed with me, and eventually, I decided to come back.

Experience working in Korea

Q. Tell us briefly about your company (size, industry focus, notable achievements).

VanaM Inc. is a startup founded in 2022, specializing in thin film deposition and operating essentially as a next-generation materials foundry. We focus on developing and commercializing new materials for emerging technologies such as quantum computing, photonics and power electronics. One of our notable achievements for 2025 is being selected as one of 12 companies out of nearly 200 applicants for Cohort 8 of the NextCorps Luminate Accelerator, a New York State funded program focused on Optics, Photonics and Imaging (OPI).

Q. How did you find your current job? Was the hiring process challenging as a foreigner?

I found my current position through Linkedin. The process was very straightforward, likely helped by the fact that VanaM is a startup. It started with an initial email exchange answering a few more/deeper questions, then an interview with the management team, and later introductions to other team members. From the first contact to joining, the whole process took about 4 weeks.

The most challenging part was getting responses in general, and the need for specialized Korean knowledge. I had to prepare extensively and review a lot of terminology (technical and others) in Korean ahead of the interviews.

Q. Can you describe your typical workday?

A typical workday follows standard office hours, roughly 9 to 6. My tasks vary widely and include investor meeting preparation, partner and international coordination, as well as marketing and strategy. We also participate in many fairs and exhibitions, most of them abroad, which require extensive planning beforehand and follow-ups afterwards.

2025 was a particularly intense year for VanaM and myself, as, as mentioned above, we were selected for the NextCoprs Luminate Accelerator Program which required monthly in-person participation in the U.S. for six months, which significantly shaped my work weeks during that period.

Q. How does Korean work culture compare to your home country's?

I would describe Korean work culture as highly disciplined and more intense than that of my home country, without implying that one is better than the other.

Korean work culture is rooted in hierarchy, a strong work ethic, and high expectations which are often met with equally strong commitment and responsibility. Korea is an extremely fast-paced country, and that inevitably leads to a demanding professional environment. This can make integration more challenging for those unfamiliar with the culture.

That said, Korean work culture is also internationally recognized for its high standards, endurance and strong sense of duty, which are qualities that have clearly contributed to Korea becoming one of the most developed countries in the world. The "Miracle of the Han River" (which refers to the incredible transformation of Korea from a war-torn country to a world leading economy in a few decades) happened for a reason!

Q. Do you speak Korean at work? How is communication handled in your team?

I have TOPIK level 5 and have completed the KIIP program, so I'm fairly fluent, which is probably one of the main reasons I got my current job. Communication at work is almost entirely in Korean, though I sometimes mix with English when the topic is particularly complex. Most of my working documents are also prepared in English.

Q. What challenges have you faced adapting to Korea's work environment?

My experience with VanaM Inc. has been incredibly positive. Like any job, there are challenges, but none of those I've personally faced there stem from the Korean work environment itself. Every member of our current Management Team has spent some time abroad for work or study, so I believe that shaped our work environment in a way that was easier for me to adapt and integrate.

Before joining VanaM though, I worked as a Research Professor in academia for about 2.5 years. That's where I encountered challenges more closely tied to Korea's broader academic work environment. Graduate schools in Korea can be extremely competitive, which sometimes leads to difficult dynamics within teams. In some cases, team members don't work on related projects, which limits help and collaboration and creates a sense of isolation. I sometimes faced (and sometimes simply witnessed) these challenges firsthand.

However, I'd like to clarify that these challenges are not limited to foreigners, though language barriers and cultural adjustment certainly make them more difficult.

Lifestyle & community

Q. What's your life outside of work like? How have you built your social life here?

When I first came to Korea as an undergrad, my social life was very much that of a student. Easily making friends, going out, having fun. When I returned in 2022 for work, I was lucky to already know people, so re-building a social life wasn't too difficult.

Most of my friends are expats around my age, so late 20s to 30s, and despite my heavy business travel schedule, we try to meet regularly.

Until this year, I was also more active in sports and exploring Korea's cultural scene (galleries, musicals, classical music…), but that slowed down significantly this year. I am hoping to get back to it soon.

Otherwise, I am your fairly typical introvert. I love reading, staying at home, resting and binge-watching shows over the weekend.

Q. How much has knowing Korean (or learning it) helped in your daily life?

100%! No doubt!

Korea, like many other countries, has its own language with its own structure and characteristics, and knowing it is essential to living here comfortably. Honestly, this applies everywhere. I don't think it's realistic to come to Korea and expect to live comfortably using only English.

English can definitely help sometimes, especially for short visits, but a long-term plan in Korea without Korean will only make your own life difficult.

Q. What was your biggest culture shock when you first arrived?

I don't think I experienced a single massive culture shock, which honestly surprised me. I even tried looking for one back in 2011, because it felt strange to move from Tunisia to Korea and not have that one defining shock. Despite the many differences, I think I was more surprised by the similarities, and perhaps that made me accept the differences with a more open mind than I expected.

That said, one thing that really stood out to me was the after-work drinking culture, how intense it could be and how mandatory it sometimes felt. That was definitely something that marked me. I know it has also become much less intense now, but at the time, it was a big surprise.

Reflections & advice

Q. Do you plan to stay in Korea long-term, and how do you see your career evolving here?

"Long-term" feels too far ahead to plan, but I can definitely see myself staying in Korea for the next few years. A big reason is my company. I really love my current job and I want to grow alongside our startup, scale it significantly, and hopefully take it to a very successful exit.

Q. What tips would you give to someone searching for a tech job in Korea?

Learn Korean! Yes, tech and science fields are more open to English compared to other sectors, but your day-to-day life will be much easier if you can communicate comfortably with your colleagues.

Technical and scientific knowledge is essential, of course, but that's something you will keep building throughout your career anyway. You'll always be learning new technologies. So, make your life easier by starting with a solid level of Korean, so you can focus your energy on the job itself. It really helps level the playing field.

Q. What can Korean companies do to better support and integrate international employees?

This might sound like a contradiction to what I've been saying all along, but being more open to using English and being a little more patient could make a huge difference. That doesn't mean everything needs to be done in English or everyone needs to speak it fluently. But it would help immensely if team leaders and colleagues were willing to give new international hires the time and space to adapt, to be patient with their evolving language skills, and to show understanding when they encounter challenges that may seem intuitive to Koreans. The best example I can think of here is how difficult it can be to write a formal or official document in Korean, even for advanced levels. In my case, I still mix English and Korean for paperwork, and get help from my colleagues sometimes.

Another helpful initiative would be a mentor-mentee system. It doesn't need to be mandatory, but a structured program that pairs willing Korean employers (or experienced foreign employees) with international newcomers would greatly support integration.

Q. What's one thing you wish you knew before moving here?

Having first moved here 15 years ago, and with Korea now feeling like home, it's honestly hard to say. Maybe I wish I knew how much university exams relied on memorization? That might have potentially saved me hours of extra studying and improved some of my grades. But at this point, I'm not even sure anymore!

Q. Would you recommend Korea to other international tech professionals, and why?

If you are willing to put in the work, absolutely. If you come with an open mind, aware that you're stepping into one of the most demanding environments with very high standards, and ready for that, then yes.

Korean tech is like the NBA All-Stars. You'd better come prepared!

Rapid fire

• Favorite Korean food: 된장찌개 (Doenjang-jjigae, a soybean paste stew)
• One must-visit place in Korea: Namsan Tower, in winter, at night. Preferably on a snowy day.
• Most surprising thing about Korean tech culture: how fast it changes!
• Favorite Korean word or phrase: 알았어! (arasso meaning "Ok! Got it!")

Connect with Sahar Ayachi

• LinkedIn: /in/ayachisahar

If you want to be next and contribute, send us an email at florian@dev-korea.com.

Ready to explore your next move? Visit Dev Korea to check the latest job openings, or if you're an employer, post a job and connect with our vibrant community of tech talent eager to contribute to Korea's innovation ecosystem.

Profile snapshot

• Name: Maxence Mauduit
• Nationality: French
• Current role & company: Product Designer and Chief Design Officer at Buzzvil
• Years in Korea: 11
• Visa status: F-6-1
• Languages spoken: French, English, conversational Korean
• Previous countries worked in (if any): France

Background & career

Q. Can you briefly introduce yourself and your current role?

Maxence: I’m an experienced Product Designer and currently serve as CDO at Buzzvil, a leading Korean AdTech company. If you use Korean apps, chances are you’ve interacted with our products without realizing it. That’s the magic, and sometimes the frustration, of working in B2B2C. As an individual contributor, my core strength is interaction design. As a leader, my goal is to amplify each team member’s impact, unlocking both personal fulfillment and collective performance.

As CDO, I ensure design acts as a connective tissue between business goals and product excellence, delivering systemic solutions that drive both operational efficiency and an outstanding customer experience.

Q. What inspired you to work in Korea?

During my studies, I spent a semester abroad at Hongik University in Seoul. Those six months had a huge impact on me. I discovered a culture and environment where I felt I could grow more than back home. After graduating, I tried to start my career in Korea but eventually failed, which ultimately pushed me to build my career in Paris first.

Q. What was your career path before moving to Korea?

I studied Interaction Design in France and completed a research master’s in Virtual Innovation at ENSAM. After graduating, I spent over three years in an R&D team in Paris, working on a real-time educational platform for professionals. Once we secured our first major client, I decided it was time to leave, saved enough to sustain myself, and moved to Korea to look for opportunities. I attended a workshop at KAIST and eventually landed my first, and current, job in Seoul.

Experience working in Korea

Q. Tell us briefly about your company (size, industry focus, notable achievements).

Buzzvil has around 130 employees today. I joined in 2014 as one of the early members when the team was about 20 people. Over the years, the business pivoted several times, from a rewarded lockscreen app to a lockscreen SDK, and eventually into a full growth solution for app publishers. The mission has always been the same: rethinking how people interact with ads.

In 2024, Buzzvil recorded ₩101.2B in revenue and ₩4.4B in profit, working with partners like KakaoBank, KakaoPay, SSG, Olive Young, LG, and Samsung.

Q. How did you find your current job? Was the hiring process challenging as a foreigner?

I moved to Korea with a six-month deadline. If I didn’t find a job, I planned to return to France. After several rejections, I applied for a Lead Designer role at Buzzvil through LinkedIn, just two weeks before my flight back.

The hiring process was smooth and fast. Two interview rounds, one focused on role and problem-solving, the other on culture fit. Everything was conducted in English, and the company already had international employees. Buzzvil sponsored my E-7 visa at the time, which took about two months. The process is much faster now.

Q. Can you describe your typical workday?

We have flexible hours. With two young kids, I usually start around 8:30 a.m. after dropping them off and finish by 5:30 p.m. Mornings are my focus time, where most of my individual work gets done. My afternoons are reserved for meetings, leadership, and operational strategy.

Q. How does the Korean work culture compare to your home country's?

Buzzvil was initially very Western, almost Silicon Valley–like. After COVID, the culture shifted as many global employees returned home. Still, overtime is avoidable, and autonomy is high.

Work in Korea tends to be more intense and metrics-driven. This is great for performance but makes it harder to prioritize initiatives where quality is harder to measure. French work culture, in contrast, allows more room for intuition and debate. As a design leader, I still feel conflicted about this balance.

Q. Do you speak Korean at work? How is communication handled in your team?

I mainly work in English. I’ve studied Korean for a couple of years after work, but I’m not comfortable leading meetings in it yet. As the company has become more Korean-speaking, this has become more challenging. Recently, AI-based translation tools and live transcription have helped significantly.

Q. What challenges have you faced adapting to Korea's work environment?

Living and traveling in Korea are very different experiences. Early on, I went through a difficult period where everything I loved about Korea became stressful. I was homesick and overwhelmed. Things improved after about a year, especially after meeting my wife. Adjusting to a new country, language, and leadership role all at once takes time.

Lifestyle & community

Q. What's your life outside of work like? How have you built your social life here?

In the early startup days, Buzzvil was my social life. We worked, traveled, and played together. As the company grew and moved toward an IPO, that naturally changed. Around the same time, we had our first child, and my life shifted toward family.

Q. How much has knowing Korean (or learning it) helped in your daily life?

Basic Korean is essential for daily life. Reading signs, ordering food, and handling errands. I focused too much on professional growth early on and regretted not investing more time in Korean sooner. I strongly recommend learning the language as early as possible, ideally before starting full-time work.

Q. What was your biggest culture shock when you first arrived?

In France, empty time is valued. In Seoul, everything is fast, dense, and always available. You can work, shop, and exercise anytime. It’s exciting, but it takes time to adjust when you settle long-term.

Reflections & advice

Q. Do you plan to stay longer in Korea, and how do you see your career evolving here?

We love it here and still feel like we’re on an adventure, even after 11 years. With two kids, we’re also aware of the challenges, especially how competitive life in Seoul can be. Buzzvil is preparing for IPO, and after years as a C-level executive, I’ve seen rapid career growth that would’ve been harder in Europe. I’m considering several future paths, but haven’t committed to one yet.

Q. What tips would you give to someone searching for a tech job in Korea?

• Be on-site. Remote applications work mainly for large companies.
• Company culture matters as much as the role. Many people leave quickly because they underestimate this.

Q. What can Korean companies do to better support and integrate international employees?

• Offer Korean language support.
• Implement buddy programs. Buzzvil has done this consistently, and it makes a real difference.

Q. What's one thing you wish you knew before moving here?

Housing prices only go up. I should have bought earlier. 🥲

Q. Would you recommend Korea to other international tech professionals, and why?

Absolutely. Seoul offers a dynamic, demanding environment. It’s not easy, but if you’re looking for growth, challenge, and constant change, it’s a place that keeps pushing you forward.

Rapid fire

• Favorite Korean food: It’s cold these days, Kimchi jjigae, hot and meaty
• One must-visit place in Korea: Oeam Folk Village near Asan, stay overnight for the sunrise
• Most surprising thing about Korean tech culture (answers below are general sentiment, not Buzzvil specific!):
  • Modern products paired with a traditional hierarchy
  • Meetings often confirm decisions rather than debate
  • Speed is prioritized, sometimes at the cost of clarity and alignment
• Favorite Korean word or phrase: 파도, it means waves. We named our first kid 파도 after spending a week near YangYang (양양). Probably the Korean word I say the most (while running after him!).

Connect with Maxence

• LinkedIn: /in/mmaxence
• Website: mmaxence.me

If you want to be next and contribute, send us an email at florian@dev-korea.com.

Ready to explore your next move? Visit Dev Korea to check the latest job openings, or if you're an employer, post a job and connect with our vibrant community of tech talent eager to contribute to Korea's innovation ecosystem.

This event marked our sixth meetup with Supabase, and the energy in the room made one thing clear. Supabase adoption in Korea keeps accelerating, and more teams are actively building and shipping with it.

From first-time users to founders and engineers already running Supabase in production, the audience reflected how widely the platform is now used across the ecosystem.

🧠 Supabase MVP essentials: scaling, security, and core features

The evening opened with a practical talk focused on building MVPs that are ready for real users.

Joseph Kim walked through the key considerations when starting a serious project with Supabase, including:

• how to think about scaling early without slowing down development
• security basics that should not be postponed
• core Supabase features that support real-world workflows

The session gave both newcomers and experienced builders a clear framework for launching quickly while staying on solid technical foundations.

🚀 Made with Supabase: community showcase

Next, the spotlight moved to the community.

Three builders took the stage to share real stories from products they built and shipped using Supabase. Instead of polished demos, the talks focused on honest decisions, trade-offs, and lessons learned while building for real users.

Each speaker walked through:

• what they built and why
• how Supabase fit their technical needs
• challenges they ran into along the way
• what they would do differently next time

Furqan Ali shared how he built Screenshow, a tool to create high-quality screen recording videos. As a solo founder, he explained how Supabase helped him move fast while keeping the stack simple and production-ready.

Ajit Kumar walked through a PostgreSQL migration from Render to Supabase for his product Random Dev. He shared the issues he encountered during the migration process, the unexpected challenges along the way, and how he resolved them step by step.

Kyubo Shim talked about migrating his app KBO Cheer Song Player from Firebase to Supabase. He highlighted the improved developer experience, simpler architecture, and meaningful cost savings he achieved after the move.

This part of the evening was especially engaging, showing three very different paths to building with Supabase, and how teams and solo builders alike can move fast, iterate, and still ship reliable products.

🌱 Supabase momentum in the Korean developer ecosystem

Compared to our first Supabase events, the conversation has clearly evolved.

This time, we heard:

• fewer questions about what Supabase is
• more discussions around architecture, performance, and production use
• more examples of teams already building and shipping with it

Being able to host six Supabase events and continue filling the room is a strong signal of the platform’s growing momentum in Korea.

🎁 Swag, questions, and community moments

After the talks, the room stayed lively and social.

Everyone picked up Supabase stickers, and we also held a raffle for Supabase T-shirts, which added a fun and relaxed moment to the evening.

Conversations naturally flowed around real use cases, shared experiences, and best practices when building with Supabase. The atmosphere stayed relaxed, curious, and welcoming throughout the night.

🍔 Burgers and hallway conversations

Food was served right at the start of the event, with burgers from CryCheeseBurger setting a relaxed and welcoming tone from the moment people arrived.

With burgers and drinks in hand, conversations started early:

• people shared what they are building
• compared technical approaches
• talked about startups, side projects, and careers
• met new faces and reconnected with familiar ones

Starting the evening with food made it easy for connections to form naturally, and those hallway conversations continued throughout the night.

🙌 Huge thanks

This event would not have been possible without:

• Supabase for supporting the event and bringing great content and swag
• Our speakers: Joseph Kim, Furqan Ali, Ajit Kumar and Kyubo Shim who shared their knowledge and experience.
• All the volunteers who helped make this event a success: Aaron McCollum, Aleksey Kim, Bertijn Pauwels, Charles Chang, Jason Yee and Kathie Lee who made everything flow seamlessly from start to finish
• All 90+ attendees who filled the venue and brought energy into the room

Seeing the venue completely full was incredibly motivating for us as organizers.

👀 What’s next

Dev Korea keeps growing, and this event made it clear that there is strong demand for practical, technical, community-driven meetups in Seoul.

We are already preparing upcoming events with global tech companies and local builders.

👉 Check out upcoming events at https://dev-korea.com/events

Want to stay connected with what we're building? Follow us on LinkedIn, X, join our Discord server, and subscribe to our Weekly Dev Korea Digest Newsletter for all the latest updates.

Here’s to more nights of learning, building, and meaningful community in Seoul. 🍻✨

Ready to explore your next move? Visit Dev Korea to check the latest job openings, or if you’re an employer, post a job and connect with our vibrant community of tech talent eager to contribute to Korea’s innovation ecosystem.

So tell me about yourself.

A simple question with endless possible answers. It is one of the most dreaded interview questions that almost every recruiter or hiring manager will ask. You may think, “Why are they asking me to walk them through my career when they can read about it in my CV?” And yes, although we may have read through your application before jumping on a call with you, it’s an opportunity for you to pitch yourself beyond your resume and for both yourself and the interviewer to see if there is alignment in 1) the experience and skills the job requires, and 2) values that would lead to positive collaboration.

I’m Soo, a tech recruiter whose career started in San Francisco after graduating from UC Berkeley. I’m currently based in South Korea and working at ChartMogul, a SaaS startup, hiring talent from around the world.

Having conducted over 1,500 initial interviews during my time at ChartMogul, I’ve seen personal introductions ranging from the candidate asking me “What would you like to know?” to telling me their whole life story. The ones who’ve successfully advanced in the interview process exhibited the same patterns during the initial screening:

(Please note that ChartMogul is a remote-first company, so I am speaking from my experience of conducting all interviews over video.)

Stay on topic

Whether you begin by talking about how your passion started at a young age or from your first professional role, be sure not to stray too far from talking about your career and stay relevant to the job that you are interviewing for.

Some candidates go off on long tangents about the complexities of a certain tool or talk in-depth about personal projects that aren’t relevant. Although we understand their intention of wanting to display their knowledge or their diverse experience, this type of verbose, irrelevant introduction can suggest to interviewers that the candidate is likely to over-explain and sidestep during team discussions.

All-inclusive summary

When walking through each of your roles, what the interviewers want to hear is enough details that give them a good picture of what you were doing that is not already written in your CV. This includes the type of product you worked on, your day-to-day responsibilities, the different projects and features you contributed to, the tools you used, the team size, and your role within the team. And if you have any short stints in your career, explaining why you left the company is a good practice.

A well-rounded summary can often answer questions that the interviewers had prepared and allow them to hone in on follow-up questions that make the best use of your interview time.

“Why are you looking for a new role?”

Is a question that most, if not all, interviewers will ask. It is a way for us to understand your motivation and to assess whether there is a long-term fit. Rather than waiting for the interviewer to bring it up, you can naturally incorporate this into your introduction by ending it with why you are looking for a new opportunity and what excites you about the job or the company you’re applying for.

The STAR method

After you’ve pitched yourself, the interviewer will ask more in-depth questions about specific projects or behavioral questions, for which you’ll be expected to provide concrete examples from past experiences. Try to structure your responses using the STAR method:

• Situation: Set the scene and give enough context, such as what company it was for, why the feature was needed, who was involved, etc.
• Task: Describe your responsibilities and any challenges you faced.
• Action: Explain the actions you took and how you carried them out.
• Result: Share the outcome of your work and highlight any quantifiable achievements.

However, we understand that it can be difficult to remember specific examples in a high-stress situation like an interview. In the case that you truly cannot remember and are unable to answer those behavioral questions, turn them into situational questions of “What would I do in that scenario?” and talk about your approach rather than sitting there stumped.

Be prepared

Whether the interview is over phone or video, treat all interviews like an in-person meeting. This means having your devices charged and updated beforehand, looking presentable rather than disheveled as if you had just gotten out of bed, ensuring that your environment is distraction-free, and practicing good body language even through the screens. For video interviews, we highly recommend taking them on a laptop or computer instead of your phone unless that’s your only option.

And if you are running late or dealing with issues that could disrupt the interview, make sure to communicate and email the interviewer ahead of time!

Most importantly: Be genuine and authentic!

Although you may think you’re getting away with it, we can tell when someone has memorized an AI-generated introduction or is reading off of a script on their screen. It’s okay to use your CV as a guide for you to follow and to practice your responses beforehand, but we want to see you be yourself and talk about your experiences naturally.

Engage in small talk, be conversational, and show the kind of person you would be as their future team member.

Here’s an example of what a strong introduction could sound like:

“I’ve been a software developer for over 4 years. I’ve mainly worked in backend with Java, but also have some experience in frontend with React.

My career started at (company name), a software service company, where I joined as a junior developer and grew to mid-level in a year. During this time, I worked with various clients in real estate, e-commerce, and other industries on their needs like building small features, working on websites, etc. I usually worked with 4-5 other developers and collaborated closely with the clients and stakeholders.

I wanted to gain experience in a product company, so I joined (company name) that has a big HRIS system and this is where I learned the most. I was part of the Tooling Team of 12 people including a PM and QA, where we handled an internal tool that was used by all the engineering teams. The biggest project I worked on was an automatic workflow process, where once a task was completed, it would automatically move on to the next step. It took about a month to complete and we shipped it without any major bugs or delays. Besides this, in my day-to-day, I was involved in mentoring new team members, writing documentation, helping with testing, and doing code reviews.

In my personal time, I’ve been trying out other languages like Python and Ruby to expand my horizons. I also like to read about different software architecture. Outside of tech, I enjoy mountain biking in my free time.

Now I’m looking for a new challenge where I’m part of a smaller team and hopefully have a bigger contribution to the company. I’d love to be able to see the direct impact I have on customers who love using the product.”

Connect with Soo

• LinkedIn: /in/y-spark
• Visit ChartMogul: jobs.chartmogul.com/

If you want to be next and contribute, send us an email to florian@dev-korea.com.

Ready to explore your next move? Visit Dev Korea to check the latest job openings, or if you're an employer, post a job and connect with our vibrant community of tech talent eager to contribute to Korea's innovation ecosystem.