# remote-in-japan: a table of Japanese remote employers

> remote-in-japan is a hand-written Markdown table of Japanese technology companies that hire remotely, kept in two language versions on GitHub. Its third column looks like a filter and is wrong often enough that the prose in each row is the only thing worth reading.

**remote-jp/remote-in-japan** — Tech companies in Japan that hire remote workers.

- Repository: https://github.com/remote-jp/remote-in-japan
- Website: https://remotework.jp
- Stars: 1,991 · Forks: 318
- Language: Ruby
- License: not declared
- Published: 2026-10-14 · Updated: 2026-10-14 · Language: en
- Canonical page: https://hysenlabs.com/projects/remote-jp-remote-in-japan

## A Markdown table is the entire product

The whole contribution of this repository sits in one Markdown table with three columns: company name, description, and full remote. Description cells are long free text written by each company, carrying links to its own careers page, its engineering blog, and its technology stack. Nothing in a row is structured, which means every use beyond skimming is a parsing job.

The project describes itself as tech companies in Japan that hire remote workers, and its homepage field points to https://remotework.jp. Repository topics are github-pages, japan, jekyll, remote-work, and ruby. The audience is narrow and obvious: an engineer or designer already in Japan, or ready to base there, weighing employers before an application rather than after one.

That comparison is the value. A recruiter site lists openings that expire in weeks, whereas this table lists companies, and a company row is the unit that survives a hiring freeze. What it cannot do is tell you whether a remote policy quoted in a cell is still true next month, because no cell carries a date.

## There is nothing to install and nothing to run

There is no install step here and the repository offers none. No package to fetch, no dependencies to resolve, no build to run: a reader opens the homepage at https://remotework.jp, or opens README.en.md in the repository for the English copy of the same table.

Before writing any selector, look at the header line you would be parsing.

```bash
| 会社名 | 説明 | フルリモート |
```

Those three headers are Japanese and mean company name, description, and full remote. A quick first use is to scan the third column for the `:ok_hand:` shortcode, then check the description cell of every row that scan dropped, for the reason the next section sets out.

After the table, the repository offers nothing except outbound links. Each row carries a recruiting page, and several add a tech blog or a department deck, which is where the real verification work starts.

## The full remote column contradicts the rows beside it

The third column holds exactly one value anywhere in the visible table, the `:ok_hand:` shortcode. Five rows carry it: kickflow, as-child, Spacely, BoostDraft, and Asobica. Seven other complete rows leave it empty, including several whose own description states a remote policy in plain words.

Leaner Technologies says its engineers and designers work fully remote from anywhere in the country, and the cell beside that sentence is blank. Greff says its engineers are fully remote with three to four office visits a year, blank as well. Inner Resource says engineers are basically fully remote with no overtime, also blank, while PORT reports that about 70% of engineers work fully remote and is blank too.

So the column records which companies volunteered a marker, not which ones allow remote work. Filter on it and you drop Leaner Technologies, Greff, Inner Resource, and PORT from a list they plainly belong to, then wonder why the results feel thin.

One row shows the mechanics in miniature.

```bash
| [エースチャイルド株式会社](https://www.as-child.com/) | ITのチカラで、子どもの未来を明るく。<br><br>エースチャイルドは、子ども達を守るために、子ども達が安心安全に育つことができる社会を実現するために、ITのチカラで明るい未来を作る会社です。<br><br>[採用情報](https://www.as-child.com/recruit/) | :ok_hand: |
```

Three fields and one short sentence: name, a recruiting link, a marker. The description is so thin that a reader learns nothing about days per week, location rules, or the team, and yet the marker is the only field in the row that can be filtered without reading the prose.

## Full remote still comes with quarterly office days

Read the descriptions and remote turns out to be a range rather than a switch. PORT lets engineers work remotely four or more days a week and says about 70% of its engineers are fully remote, which is a policy and a workforce statistic in one sentence.

Greff calls its engineer roles fully remote and then adds office attendance three to four times a year. Inner Resource is more careful with the wording: engineers are basically fully remote, there is no overtime, and anyone living inside the commuting radius may be asked in about once every two weeks. Leaner Technologies goes furthest and says engineers and designers work remotely from all over the country.

BoostDraft stretches the idea past Japan entirely. Its row describes a fully remote structure with members from different countries, paired with an annual offsite, an optional office day each month, and recurring habits such as Morning-checkin, Game Day, and Coffee-chat. Optional in the row does not mean infrequent, and a reader working from the marker column never learns which of those three shapes an employer actually offers.

The practical consequence is that remote work in Japan covers a job with no commute at all, a job with one office day a fortnight, and a job that asks for four trips a year. No single column separates the three.

## The stack lines are the real search index

Every description carries a technology line, and that is the part of this table worth treating as structured data. Leaner Technologies lists Ruby, Rails, JavaScript, TypeScript, React, Next.js, Storybook, MySQL, and AWS for a procurement SaaS.

Greff lists PHP, Laravel, TypeScript, Next.js, Flutter, and AWS, with no Ruby anywhere in it, which tells you at once that this is not a Ruby job board. kickflow sits at the other end: Ruby on Rails, RSpec, Rubocop, and Sidekiq on the backend, with HTML, CSS, TypeScript, Vue.js, Nuxt.js, Vitest, and ESLint on the front end, and a row that pairs that stack with full remote and full flex.

Canary adds Go, Next.js, React Native, and GCP, then says GitHub Copilot, Cursor, and Devin are available, that Claude Code is in trial operation as of 2025/6, and that every employee holds a paid Gemini plan. Spacely and AP Communications each link an engineering blog, so two rows let a reader check working style long before any interview.

For a reader who already has a CV and a filter, that line is the only part of a row that matches cleanly against what is known.

## Every size number is company recruiting copy

Every figure in the table comes from the employer describing itself, and it should be read that way. Canary reports 500万DL突破 for its consumer app alongside an App Store category rating of 4.8, and 累計利用者数200万人突破 for its B2B SaaS, on a staff of roughly 120 people in a growth phase aiming at a listing.

Fenrir claims 400 companies and more than 600 apps built. Spacely claims adoption by more than 10,000 users and holds several patents covering its 360 and 3D pipeline. Inner Resource sells reprua, a purchasing, inventory, and equipment service for research labs, and its row leans on autonomy for a small team instead of on scale at all.

AP Communications is the outlier, because it publishes a workforce number: as of January 2025, slightly under a fifth of its staff live outside the prefecture and work remotely. That makes it the only row here that quantifies remote headcount directly, and the figure still originates with the employer rather than with any independent check.

Use these numbers for one purpose, judging whether a team is large enough to have a real engineering process or small enough that you would own a whole surface, and take the rest as marketing.

## Breadth against freshness, and the company page wins on dates

Breadth is what this repository has and freshness is what it does not. One table hands over a dozen employers at once, each with a linked recruiting page, which is more than most readers assemble in an afternoon of searching.

The alternative that beats it is the obvious one. Open the recruiting page the row already links, where the employer writes about current openings in its own words, and you get the policy as it stands today instead of as it stood when somebody wrote the row. That page offers no comparison whatsoever, since every employer describes itself as the ideal workplace, and closing that gap is exactly what the table is for.

One row shows how the dates drift. Canary states that Claude Code is in trial operation as of 2025/6, while the last push to this repository was on 2026-10-03. Nothing marks which rows have been rechecked, and the repository states no rule for adding, editing, or removing an entry, no issue template, and no schema for a row. A row two days old therefore looks exactly like a row two years old.

## README.md in Japanese and README.en.md as the copy

The primary file is README.md and it is written entirely in Japanese, table headers and description cells alike. The English version is a separate file, README.en.md, linked from the top of the Japanese one, and those two files are the only content entries at the top level.

```bash
.gitattributes
.github/
.gitignore
README.en.md
README.md
docs/
```

Two hand-maintained copies of one table means two edits for a single correction, and the English text is a translation of copy each company wrote for its own site. Stack names, product metrics, and remote wording all have to survive that translation, and nothing checks whether the two files still agree with each other.

The remaining entries add little. Topics claim jekyll and github-pages, the primary language is Ruby, and the homepage is https://remotework.jp, yet no Jekyll configuration file appears among the six top-level entries, so the route from this Markdown table to that site cannot be followed from the repository listing. The docs directory is present and undescribed, and .github holds repository settings a reader never sees.

Which leaves an honest and small use for the project: it saves the work of assembling a shortlist by hand, and it claims nothing about any individual row being current.

## Conclusion

Treat remote-in-japan as a shortlist generator, not a source of truth. Engineers and designers in Japan who read Japanese get a dozen employers, the stack line in each row, and the prose remote policy that the third column hides. Anyone needing a filterable feed, a reliable full remote flag, or a hiring date has to look past it. Before applying, open the recruiting page each row links and read the remote terms there, because no cell carries a date and the Canary row still dates its tooling to 2025/6.

## FAQ

### What is remote-in-japan and who is it for?

It is a Markdown table of Japanese technology companies that allow remote work, kept in a GitHub repository with a homepage at https://remotework.jp. It suits engineers and designers already in Japan who are weighing employers before applying.

### How can I tell whether a company on the list is fully remote?

Read the description cell rather than the third column. Only five of the twelve complete rows carry the `:ok_hand:` shortcode, and companies such as Leaner Technologies, Greff, Inner Resource, and PORT state their remote policy in prose while leaving that column empty.

### Does the repository offer a filterable feed or an API?

No feed, schema, or script is listed among the top-level entries, which are two READMEs, a docs directory, and repository settings files. Filtering work therefore means parsing the Markdown table yourself and treating the description column as text.

### Is remote work in Japan the same thing at every company listed?

It is not. PORT allows four or more remote days a week with about 70% of engineers fully remote, Greff pairs full remote with three to four office visits a year, and Inner Resource may ask staff inside the commuting radius in about once every two weeks.

### How current is the list?

The last push was on 2026-10-03 and the repository is not archived, but individual rows carry no dates at all. The Canary row dates its AI tooling to 2025/6, so each company page is worth opening before a row is relied on.

## Sources

- [Issues](https://github.com/remote-jp/remote-in-japan/issues)
- [Project website](https://remotework.jp)
- [README](https://github.com/remote-jp/remote-in-japan/blob/main/README.md)
- [remote-jp/remote-in-japan on GitHub](https://github.com/remote-jp/remote-in-japan)

---

Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/remote-jp-remote-in-japan