@artscoder·1d24 platforms pay people to train AI. We rated them on pay and trust.
Median published pay: ~$20/hr floor, ~$60/hr ceiling. Specialists earn far more.
Red flag #1: an "AI job" that asks YOU to pay.
Full ranking + scam checklist: https://remoty.work/ai-training-jobs/?utm_source=whop&utm_medium=social&utm_campaign=ai-training-jobs
@artscoder·Oct 1New: The Remote Job Board Index.
We measured 14 remote job boards for 30 days instead of reviewing them — 9,994 live listings across 4,508 companies.
85.4% of NoDesk listings arrive with no usable description. 59.7% on 4 Day Week. A title, a company, and nothing describing the work.
And pay transparency splits into two different questions. Jobicy carries pay information in 57.9% of its listings but exposes it in a machine-readable field only 1.1% of the time. The salary is there — it's buried in prose, so no filter can reach it.
Each board now has its own page with its own numbers, refreshed monthly:
https://remoty.work/job-boards/?utm_source=whop&utm_medium=social&utm_campaign=board-index-oct
@artscoder·Sep 21We deleted 1,650 jobs from our own board today. Not because they expired — because they should never have been there.
1,556 had no description at all. A job title, a company name, an empty page. Our own scoring model had already flagged them: asked "would a candidate have enough here to decide whether to apply?", it scored these 0.06 out of 1, against 0.66–0.92 for everything else. One source had been storing titles without bodies for months. The adapter is fixed; the old ones are unrecoverable because those jobs have since been delisted upstream.
The other 94 were office jobs. On a remote job board. San Francisco, New York, London. Not mis-tagged either: 91 of the 94 never use the word "remote" anywhere in the posting, 15 say on-site outright, 7 say hybrid.
10,162 listings → 8,500. Every count on the site dropped with it automatically, because they all read from one rule.
A smaller number you can trust beats a bigger one you can't. If a board's headline count is what it optimises, the quiet way to grow it is to stop checking.
@artscoder·Sep 15Yesterday our board said 119 of 10,336 live listings published a salary. About 1%. Great stat. Wrong stat.
Our biggest source is roughly half the board — about 5,200 live listings — and exactly one of them carried a salary. Not employer secrecy: the API sends minimum, maximum, currency and period, and our importer hardcoded the field empty months ago. Nobody caught it, because "employers hide pay" is an entirely believable number.
Fixed. Of everything ingested in the 24 hours since, 25% publishes pay. The board overall still reads 3.7%, because listings imported before the fix keep the blank field until they age out of the 30-day window — so it climbs on its own over the next month.
The lesson is not about salaries. Missing data on an aggregator is evidence about the parser, not about the market — and "the market doesn't do X" is the most comfortable possible explanation for "we failed to read X".
Filter for pay-published listings: https://remoty.work/jobs/?salary=1
@artscoder·Sep 14We counted every remote opening on the board from the last 30 days: 10,288 listings across 4,321 companies.
2,828 of those companies posted exactly one role — 65% of the employers hiring remotely right now, and 27% of the whole board. The ten biggest hirers are only 8% of listings; the top fifty, 17%.
Remote hiring is a long tail of small teams with one seat to fill, not a handful of big names. That is where the odds are: a recognisable company collects thousands of applicants per role, a 25-person team that posted once this month collects dozens — and usually replies.
Caveats, because they matter: this counts the boards we scrape, not all remote hiring, and a few of the biggest "employers" in the top ten are talent marketplaces posting bench roles, not one company staffing up.
Measured 14 Sep 2026. Filter it yourself: https://remoty.work/jobs/
@artscoder·Sep 6We audited our own Apply buttons, and the honest result is not flattering.
10,188 remote jobs are open on remoty.work right now, across 3,917 companies. Every one has an Apply button. Where does it actually go?
347 verified — redirect followed, ends on the employer's own application page or ATS.
317 traced — earliest sighting points at a non-aggregator source, not re-checked, so not called verified.
9,524 unproven — opens the link we were handed, very often another job board.
That last number is the ordinary state of every job aggregator on the internet, ours included. The difference is that we print it. Each listing names the host the Apply button opens, says which tier it is in, and shows the full sourcing trail: every place we saw that posting, when, and from which source.
Two findings worth passing on:
The biggest gain was not more redirect-following. One of our sources had been storing employer ATS URLs in a field we were ignoring — that single fallback resolved about three times what six weeks of redirect-chasing had.
Redirect resolution decays. A link verified in July is not verified today, which is why the verified count over the live 30-day window is far below our all-time resolved count. Quoting the all-time figure as if it described the open board would have been the flattering, wrong move.
Board: https://remoty.work/jobs/
Measured 6 September 2026, over listings open in the last 30 days.
@artscoder·Sep 1AI job market, from inside the database: 52 AI-titled listings open right now. 8 are talent-network ads (a signup, not a role). 13 carry ghost-job signals — the same posting cycling back, or open far longer than the role could plausibly stay unfilled.
The ones actually paying, live today:
AI Engineer, GTM — Anthropic (graded A): https://remoty.work/jobs/44902/
GTM AI Engineer — Customer.io, $140,000–$170,000: https://remoty.work/jobs/44408/
Remote AI Trainer (coding/finance/medicine) — TrainBrainSpace, $45–$100/hr: https://remoty.work/jobs/42030/
AI Data Annotator — Argos Multilingual: https://remoty.work/jobs/46170/
Before applying anywhere in this market: check how long the posting has been up and whether the same title keeps coming back. Every listing page now shows that history — first board, every repost, days on the market.
@artscoder·Aug 31This week's index numbers, across 1,016 companies:
2,465 open remote listings
378 disclose a salary (15%)
294 flagged as possible ghost jobs (12%)
495 genuinely worldwide (20%)
15% salary disclosure is the number worth sitting with. The ones that do publish a range are not all giants either — Databricks at $308,700 and Gong at $240k-$300k are in there, but so are two-person startups. It is a choice, not a company-size constraint.
Caveat, because it matters: 15% is where we can read a range off the listing itself. Buried in a PDF or "we'll discuss on the call" scores the same as no disclosure, because to you it is the same.
@artscoder·Aug 29Graded listing of the day: Multigres Engineer at Supabase, worldwide. Score B.
Checked against ghost-job and repost patterns before it made the board.
@artscoder·Aug 25Graded listing of the day: GTM AI Engineer at Customer.io, $140,000-$170,000, posted today. Score 80/100 (B).
+25 pay band published, +20 real scope, +15 fresh, +15 remote clarity, +5 corroboration (one source only), +0 skill tags.
Watch-out: listing header says "Anywhere in the World", description says "Americas Remote". Ask about timezone before applying.
