Job and bounty data, without the noise

Two small tools. Both throw away most of what they download and tell you exactly what they threw away.

Run on Apify · public sample data · pay per result · last measured 2026-10-24
How we chose these two products (55 niches measured)

New: how old are the still-"open" postings? — median 51 days, 14.2% older than six months, oldest 2,628 days.

What we sell

259 companies bundled · 6 ATS connectors

ATS Job Feed

$0.0015 / job + $0.0005 / run

Pulls open roles straight from the applicant-tracking systems companies actually use — Greenhouse, Lever, Ashby, Workable, Personio and Workday — then normalises every pay range to annual USD, removes duplicates across boards, and can return only postings that state a salary.

Measured, whole bundled pool	Count
Companies verified reachable	259
Open postings scraped	18,825
Postings that state a pay range	8,071
…after cross-board de-duplication	6,848
Cost of one full-pool run (paying customers)	≈ $0.11

Salary coverage is honest and uneven: Ashby publishes structured pay on ~52% of postings, Greenhouse ~44% (we read it out of the job description), Lever ~9%. Workable publishes none. If a posting has no salary we leave the field empty — we never estimate it.

Open on Apify → Sample JSON Sample CSV See it as a table

5 sources · noise removed

Freelance Lead Radar

$0.015 / lead + $0.005 / run

Scans GitHub bounties, Hacker News contract threads, Freelancer.com and YesWeHack bug-bounty programmes, drops the for-hire posts and the bait repositories, estimates effort in hours and ranks every lead by USD per hour instead of headline size.

Why the filter exists	Measured
Open GitHub issues tagged `bounty`	4,640
…of the newest 200, how many name a dollar amount	4
…of those 200, how many came from one spam repo	191
Median bounty in a 298-post sample of `💎 Bounty`	$5
Top-level comments in the October HN “Freelancer? Seeking freelancer?” thread that were job offers	0 / 21

Open on Apify → Sample JSON See it as a table

Look before you pay

Both sample datasets above are real runs with default input, readable without an account. They are also mirrored as static snapshots on this site, so they stay readable even if the Apify dataset is cleaned up later. One job record looks like this:

{
  "company": "openai",
  "ats": "ashby",
  "title": "Research Scientist / Research Engineer - Foundations, Search",
  "location": "San Francisco, CA",
  "remote": false,
  "salaryLow": 380000,
  "salaryHigh": 850000,
  "salaryCurrency": "USD",
  "salaryUsdLow": 380000,
  "salaryUsdHigh": 850000,
  "postedAt": "2025-06-16T22:35:00.347000+00:00",
  "url": "https://jobs.ashbyhq.com/openai/…"
}

Free dataset: published salary ranges

Mirror on GitHub (same CSV, CC BY 4.0): sjdfi0wjvosdju9/job-salary-open-data.

While building the job feed we ended up with something useful on its own — a census of what companies actually publish as pay. 6,823 postings with an annual range, from 259 companies, normalised to USD, median midpoint $204,750. The CSV is free, no signup, no email: salary-data.html · salary-ranges.csv.

Not sure which company? run it · Type a company name and see the bands it publishes.

Per-company pages (145 of them) live under /pay/ — median, p25/p75, role breakdown, and the highest published ranges for each company.

What we deliberately do not do

We never submit bids, applications, or bug-bounty reports. The tools read public listings; you do the sending.
We never guess a salary that a posting does not state, and we drop implausible ones rather than publish them.
We do not promise a posting is still open by the time you act on it.
No browser automation, no proxies, no captcha solving. Every source is a public, keyless endpoint.

Notes on how it breaks

Job boards rotate their slugs. When a company vanishes we report it in failedCompanies instead of silently returning less.
Workday needs the full careers URL — its board-host prefix cannot be guessed reliably; 192 guesses produced exactly one hit.
Bug-bounty programmes publish a reward range. For ranking we use the geometric mean of the range, not the ceiling.

Writing

We publish the measurements behind these numbers, including the mistakes: I built a job-lead radar because the job boards I tried returned garbage, what 259 companies actually publish as pay, and why remote engineering roles publish 11.5% less on dev.to.