DEV Community

Cover image for 715 tech jobs in Korea, straight from company job boards
seoulsmith
seoulsmith

Posted on Originally published at seoulsmith.com

715 tech jobs in Korea, straight from company job boards

Originally published on seoulsmith.com.

Horizontal bar chart of 715 tech job postings located in Korea on November 9, 2026, by company: Coupang 350, 42dot 78, Krafton 67, FuriosaAI 53, Daangn (Karrot) 48, Databricks 18, Anthropic 14, Datadog 14, Moloco 10, OpenAI 9, Riot Games 8, Binance 5, and 41 at 16 other companies.
Collected on November 9, 2026 from the public job board APIs of Greenhouse, Lever and Ashby. Coupang alone has almost half of the postings.

A week ago I launched my first Apify Actor, a tool for Korean apartment prices. Right before launch I found at least seven other Actors built on the same government dataset, most with about one monthly user. The niche was crowded with builders, not customers.

So for the second one, I changed the order. I looked for demand first and wrote code last. This post is about what I found, what I left out, and one bug that would have quietly ruined the data.

Starting from demand this time

Job data is one of the few categories on Apify Store with clear, repeat demand. When I checked on November 9, the most popular Actor that reads company job boards had more than 1,600 users. Four newer Actors in the same category, all launched this year, had reached 48 to 137 users within three or four months.

Then I searched the Store for Korean jobs. There were already about 30 Actors for Korean job sites, the big local platforms where most Korean postings live. Most had between 1 and 36 users.

None of them used the companies' own job boards. That was the gap I could see.

What I left out, and why

The obvious move would have been another scraper for a big Korean job site. I didn't build one, for two reasons.

First, it's the crowded part. Thirty Actors on the same few sites is the same situation I had just walked into with apartment prices.

Second, the terms. One Korean hiring platform's terms strictly prohibit large-scale use such as crawling and mirroring. Another one's terms forbid using the service for commercial purposes without consent. That clause is written for members and needs interpretation, so I left it out until I can ask them directly. I have a day job, and my rule for this side project is simple: the legal risk has to be something I could explain calmly to my employer.

Company job boards are different. Many companies post jobs through an applicant tracking system (ATS) like Greenhouse, Lever or Ashby, and these systems offer public, unauthenticated APIs for exactly this purpose: showing a company's open jobs on other sites. Lever's documentation says it plainly: "These jobs may be scraped by third parties." I also checked each host's robots.txt before sending a single request.

Finding the companies

The hard part wasn't calling the APIs. It was knowing which companies to call. A Greenhouse board lives at a name like coupang or krafton, and there is no public list.

So I built a small discovery script. It started from 104 candidate companies (54 Korean, 50 global companies likely to hire in Korea), tried 138 possible board names on all three systems, and checked whether each board had any postings located in Korea:

  • 414 name-and-system combinations, 454 requests in total
  • At most one request per host every 1.1 seconds
  • Zero blocks or errors
  • 54 boards found, and 28 of them had postings in Korea

The miss list was interesting too. I couldn't find boards on these three systems for several of the best-known Korean tech companies, including Toss, Baemin, Kakao, Naver and Nexon. My guess is that they run their own hiring systems, but I haven't confirmed that.

The bug that put Singapore jobs in Seoul

My first full run returned 731 postings "in Korea." Some of them were not.

Greenhouse can return a list of offices for each posting. For global companies, that list often includes every office in the region. A job in Singapore or Hong Kong would list Seoul as one of many offices, and my filter happily counted it as a Korean job. I found real cases at Roblox and Binance.

The fix was to trust the location shown on the posting itself, and to use the office list only when that location is empty. I also excluded two false friends that a naive "Korea" match catches: Koreatown and North Korea. After the fix, the count matched what the discovery script had found, plus one posting that went up that day.

Nothing crashed, and every row looked fine on its own. It was only wrong in aggregate, which is the most dangerous kind of wrong for a dataset.

What the data shows

The first public run, on November 9:

  • 715 postings located in Korea, at 28 companies
  • 27.6 seconds, 31 API requests, no warnings
  • Platform cost: $0.0044, with a peak of 133 MB of memory (so I set the default to 256 MB)

Coupang dominates, with 350 of the 715 postings. The rest are spread across Korean companies like 42dot, Krafton, FuriosaAI and Daangn, and global companies with teams in Korea like Databricks, Anthropic and Datadog. Seoul has 585 postings and Pangyo 81. The other 49 don't name a city.

Horizontal bar chart of the 715 postings grouped by job title keywords on November 9, 2026: software engineering 231, product or program 91, AI or ML 74, sales or business 54, data 35, design 19. A posting can match more than one group.
Postings grouped by keywords in the job title, November 9, 2026

A few things surprised me:

  • Half the postings have Korean titles: 357 of 715. If you search only in English, you miss about half of them.
  • Only 1 of the 715 postings has a salary in the official salary field.
  • 77 postings were new in the last 7 days. Coupang added 46 and Krafton 11.

Every row links to the original posting, so you apply on the company's own page.

Where it stands

The Actor is live and free: Korea Tech Jobs in English.

Honest numbers: revenue $0. The Store shows a couple of runs from other accounts, but Apify runs automated daily tests on every public Actor, and the timing matches those tests, so I count real external users as zero for now.

That's why the distribution plan matters more than the code this time. The Actor runs every Monday morning and saves a weekly snapshot, and every Tuesday I'll post one number from it on X. Once a month I'll write a short report here.

If you're hiring in Korea, or looking for a job here and can't read every Korean posting: what would you want from this data first? Salary is clearly missing. What else? You can find me on X at @seoulsmith_dev.

Top comments (1)

Collapse
 
wbcomdesigns profile image
Wbcom Designs 💡 •

The gap we see is smaller companies that don't use an ATS at all and post jobs on their own WordPress careers page. Those are invisible to API collectors unless the page has JobPosting structured data. Did you find many companies you couldn't reach because they had no ATS? Disclosure: we're Wbcom Designs and make a WordPress job board plugin, so we're curious from that side.