One dataset from Greenhouse, Ashby and Lever: a clean way to track job postings

#webscraping #api #career #automation

Every applicant tracking system speaks a different dialect

If you track hiring signals, you have probably written this loop more than once:

Greenhouse returns jobs[].content as double-encoded HTML, and salary sometimes hides in a metadata array of label/value pairs.
Ashby puts structured pay on compensation.summaryComponents, not on the tier object everyone grabs first — so a naive parser silently returns nulls.
Lever has no salary field at all on its public endpoint, and sets isRemote: true on some hybrid roles.
SmartRecruiters serves descriptions from a per-posting endpoint only.

So you end up with four parsers, four schemas, and four sets of bugs.

One schema instead

The ATS Job Scraper normalises Greenhouse, Ashby, Lever and SmartRecruiters into a single row per job:

{
  "jobUid": "greenhouse:databricks:7712345",
  "source": "greenhouse",
  "companyName": "Databricks",
  "title": "Senior Data Engineer",
  "department": "Engineering",
  "locations": ["San Francisco, CA", "Remote - US"],
  "workplaceType": "HYBRID",
  "isRemote": false,
  "postedAt": "2026-09-18T09:14:02.000Z",
  "ageDays": 14,
  "applyUrl": "https://boards.greenhouse.io/databricks/jobs/7712345",
  "salary": {
    "minAmount": 180000,
    "maxAmount": 230000,
    "currency": "USD",
    "interval": "YEAR",
    "evidence": "ats-field"
  },
  "recordHash": "9bd2c1ea77405f38"
}

The field that matters most is salary.evidence. It is ats-field only when the ATS published a compensation field, description-text when a figure was quoted from the posting body, and none otherwise. Numbers are parsed from published figures, never invented — which is rarer than it should be.

Run it in 30 seconds

The main input is an array, so one run covers many employers:

{
  "boards": ["greenhouse:airtable", "https://jobs.ashbyhq.com/ramp", "lever:spotify"],
  "maxJobsPerBoard": 100,
  "maxJobsTotal": 500
}

Or from the API:

curl -X POST "https://api.apify.com/v2/acts/axiorasolutions~ats-job-scraper/runs?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"boards":["greenhouse:airtable","ashby:Linear"]}'

Bare tokens are auto-detected across providers, and a careers URL is accepted as-is.

Keep it fresh

jobUid is stable across runs, and recordHash changes when the title, locations or description change. Run it on a schedule and diff the two fields to get a precise "what is new / edited / gone" feed without storing the whole dataset twice.

A new "Head of Data" posting is a buying signal. A reposted role with an edited description is a budget signal. Both are one diff away.

Try it

Use it directly in the Apify Store — no API key, no login: ATS Job Scraper. Set maxJobsPerBoard to 5 to test it for a few cents.

DEV Community