# dev-conf-replay: a link index for Korean conference recordings

> One README, no code, no releases, and a scheduled workflow that checks whether any of several hundred video links still resolve.

**hibuz/dev-conf-replay** — 🍀 최근 국내 IT 세미나 및 개발자💻  컨퍼런스 영상의 다시 보기👀 링크를 한곳에 정리했습니다!

- Repository: https://github.com/hibuz/dev-conf-replay
- Stars: 1,163 · Forks: 54
- Language: Unknown
- License: not declared
- Published: 2026-10-06 · Updated: 2026-10-06 · Language: en
- Canonical page: https://hysenlabs.com/projects/hibuz-dev-conf-replay

## A repository that is one file and a link checker

The entire project fits in two paths: `README.md` and a `.github/` directory. There is no source code, no package manifest, no build system and no release history. The GitHub language breakdown is empty because there is no language to detect, and the repository description is a single line of Korean announcing a collection of domestic IT seminars and developer conference replays.

What sits in `.github/` is the part that turns a bookmark list into a maintained project. A workflow named `check_links.yml` runs against the README and reports through a status badge at the top of the file. For a resource whose only value is that its links work, that is the correct amount of automation: the list cannot rot silently, because a dead link produces a failing check rather than a broken promise discovered months later.

Two other badges sit alongside it. One points at the repository's commit feed in Atom format, so the list can be followed as a feed rather than watched. The other is a hit counter pointed at a separate GitHub Pages username, which suggests the project started life as a small web page and was folded into a repository.

The last push to the repository was 2026-09-09, and there are no open issues. For a document rather than a codebase, that combination is about as good as it gets.

## Newest entries first, because a list ages from the top

The README opens with a small block of recent additions, each carrying a date, a category tag, the organiser and a playlist link. The three at the time of writing run from 2026.08.26 through 2026.07.21, and they cover a .NET community conference in Busan, a big data conference and a Unity Korea event in Seoul.

Putting that block above everything else is a small design decision with a real payoff. A chronological link list is worst at the top, where a year of dead conference pages accumulates, and best at the bottom, where this week's recording was just published. Inverting it means the first thing a returning visitor sees is the freshest material, and the top of the file stays the part most likely to resolve.

The category tags on each entry are anchors into the topic sections further down, so a new addition also advertises where in the taxonomy it belongs. That matters for a list that has grown past a hundred entries, because it is what keeps the file navigable rather than turning it into a wall of links.

Below the recent block comes a quick-jump row of anchor links to every section, which on a long README is the difference between usable and annoying.

## Company sections carry the engineering detail

The largest part of the README is grouped by employer, and this is where the project earns its keep. Each company entry lists its recurring events with per-year playlist links, so you can watch one conference rather than hunt through a channel.

Naver appears with several distinct series: a developer conference running since 2020, an internal team conference split by year, an engineering day, an older tech concert, and a search-specific colloquium. Kakao has its annual developer conference with per-day playlists for recent years and a separate tech meetup series numbered by edition. LINE has a developer day with entries from 2020 through 2026 plus a developer meetup. The delivery platform has its annual conference across six editions, a monthly tech seminar, and two smaller community events. The local marketplace company runs a tech meetup split by discipline, with SRE, server, machine learning, database and platform tracks as separate playlists, and its payments team publishes a separate livestream.

The fintech company rounds out the section with a makers conference that splits into engineering, design and product days, a multi-year slash event, and a design-focused conference with irregular years.

The pattern is consistent: one event per year, one link per year, and a date at the end when the exact day is known. Where the format loosens, it loosens because the source loosened, which is how you can tell a playlist someone uploaded after the fact from a channel the company maintains.

## Fifteen topic sections underneath the companies

Below the company sections sits a topic index, and the quick-jump row at the top names all fifteen of them. They run from IT companies and artificial intelligence through big data, cloud, infrastructure and DevOps, blockchain, mobility, games, security, mobile, frontend and JavaScript, programming languages, open source, education and community, ending with an other section for everything that does not fit.

That taxonomy is the second half of the navigation problem. Company grouping answers who spoke, topic grouping answers what about, and a reader looking for a Kubernetes talk does not know which company gave it. Having both means the same recording is reachable either way, which is the difference between a list people bookmark and a list people search.

The categories are also a reasonable sketch of what the Korean developer conference scene actually covers. Infrastructure and DevOps and cloud appear as separate categories, which says something about how much of the indexed content is operational rather than product work. Programming languages has its own section, and the most recent addition in the newest block is a .NET community event filed under it.

Having an other section is the detail that suggests the taxonomy is applied honestly rather than for show. Every classification scheme needs an escape hatch, and a list that pretends otherwise ends up with entries filed where they do not belong.

## A ranked channel list with its own update date

Between the recent additions and the company sections sits a separate ranked list of developer YouTube channels, ten entries ordered by subscriber count, with the counts written out and a date attached to the section itself.

The date matters. The section states it was last updated on 2026.07.17, which is a different kind of promise from the rest of the README. The conference links are appended as they appear, so the top of the file always looks current, whereas a ranked channel list is only as true as its last review. Stamping the date lets a reader judge staleness instead of assuming it.

Subscriber counts are written in a compact Korean numeric convention where the unit is ten thousand, so the top entry reads as roughly 744,000 rather than 74.4 million. That formatting detail is small and it is also the kind of thing that makes a translated README slightly misleading if you read the numbers literally, so the conversion is worth doing before quoting any figure from that block.

The list itself is a different resource from the conference index, closer to a reading list of channels that publish technical material in Korean. It is also the part most likely to drift, since channel popularity and channel survival both change faster than a conference archive does.

## What this project is not

It is worth being clear about the boundaries, because the shape invites assumptions that do not hold.

It is not a searchable database. There is no API, no index file, no tags beyond the section headings, and no filtering. Finding something means reading headings or using the browser's find function, which is a real limitation once the file is long enough that scrolling is slower than searching.

It is not a transcript archive. Every entry is a link to a playlist or a video on someone else's platform, so availability depends entirely on the uploader. Nothing here is mirrored, re-encoded or backed up.

It is not multilingual by intent. The scope is Korean domestic seminars and conferences, which in practice means Korean-language sessions from Korean employers. An English talk from a Korean company may appear by accident; an English talk from anywhere else will not.

And it is not licensed, which for a document nobody intends to reuse as a work product is normal, but it does mean you should link back rather than republish wholesale.

None of this is a criticism. A single README with a link checker and a recent-first ordering is exactly the right amount of machinery for the job, and the 1163 stars the repository has collected reflect how often people come back to it.

## Conclusion

dev-conf-replay is the right resource if you read Korean and you would rather not hunt through a company's YouTube channel for the engineering talk you half remember. It is a link index with no code to install, no release cadence to track and no licence file, so adopting it means watching, subscribing or bookmarking rather than depending on it. Two limits are worth naming before you rely on it. It is scoped to Korean-language sessions from Korean companies and conferences, so English-language material is essentially absent by design rather than by omission. And a link list ages quickly, which is why the repository runs a link check workflow and why zero open issues is not the same as zero dead links. Start from the newest entries block, which carries the three most recent additions with their dates, since that is the section most likely to still resolve, then check the channel list before spending an evening on a playlist.

## FAQ

### What does dev-conf-replay contain?

It is a single README indexing replay links for Korean IT seminars and developer conferences. Entries are grouped first by employer, with recurring events broken out year by year, and then by fifteen topic sections covering AI, big data, cloud, infrastructure and DevOps, blockchain, mobility, games, security, mobile, frontend and JavaScript, programming languages, open source, education and community. A block of recent additions with dates sits at the top of the file.

### Is the list only in Korean?

The scope is Korean domestic seminars and conferences, so the recordings are predominantly Korean-language sessions from Korean employers, and the README itself is written in Korean. An English-language talk may appear where a company chose to publish one, but English material from other countries is not what the list is collecting. The channel section adds ten Korean developer YouTube channels ranked by subscriber count.

### How are dead links handled?

A GitHub Actions workflow defined in `.github/workflows/check_links.yml` checks the links in the README and reports through a status badge at the top of the file, so a broken link surfaces as a failing check rather than as a broken link a reader discovers later. The repository has a commit feed badge as well, which lets you follow additions without watching the page. Nothing is mirrored or re-hosted, so availability still depends on the original uploader.

### How current is the list?

The repository was last pushed on 2026-09-09. The most recent entries in the additions block run from 2026.08.26 through 2026.07.21, covering a .NET community conference, a big data conference and a Unity Korea event. The ranked developer channel section carries its own update date of 2026.07.17, since a popularity ranking goes stale faster than a conference archive does.

## Sources

- [hibuz/dev-conf-replay on GitHub](https://github.com/hibuz/dev-conf-replay)
- [Issues](https://github.com/hibuz/dev-conf-replay/issues)
- [README](https://github.com/hibuz/dev-conf-replay/blob/main/README.md)

---

Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/hibuz-dev-conf-replay