Methodology
How listings work
This page is the editorial policy for the board. If a listing looks wrong, start here, then tell us.
Where jobs come from
Openings are maintained in a structured feed (company, title, pay range, apply URL, location, tags, posted note). The site reads that feed on the server and regenerates job URLs about every five minutes. We focus on roles that are actually data annotation, RLHF, or evaluation work, not every job a lab happens to post.
Company profiles on the companies index are written by us from public information (sites, news, filings, and industry estimates). Revenue and valuation figures are estimates, not audited financials.
What we do not promise
- We do not confirm that every apply link is still accepting candidates the moment you click it. Employers open and close pipelines without notice.
- We do not run background checks, skills tests, or payroll. Those belong to the company on the apply button.
- Pay ranges are copied from the listing or from a public post. They are not an offer. See the salary guide for how to read them.
- “Posted today” style dates are as supplied in the feed, not a crawl timestamp we invent.
Corrections and additions
If you hire annotators and want a role listed, or you are a worker who spotted a dead link, use contact. Include the company name, title, and apply URL. We will not publish unpaid “featured employer” placements as if they were organic results.