ATS Tech Stack Detector — Greenhouse, Lever & Ashby

$5.05 / 1 000 rows  ·  pay only for results  ·  no credit card to try

We do the dirty work so your dataset stays clean. 😈

B2B tech-stack intel from public job posts — no SaaS subscription. Point this Actor at any company's Greenhouse, Lever, or Ashby board and get one structured row per active job, including a deduplicated list of canonical tech names (Postgres, Django, AWS, Kubernetes…) detected directly from the job description.

Sales reps qualifying accounts, recruiters sourcing by skill, and competitive-intelligence analysts mapping vendor footprint all need this signal. Commercial tools (TheirStack, Wappalyzer, BuiltWith) charge subscription fees for the same data — and they sniff front-end JS, not the back-end stack. This ATS tech stack detector reads the job descriptions where engineering teams declare their actual infrastructure.

🎯 What this scrapes

Three ATS platforms, one unified schema:

1. Greenhouse — boards-api.greenhouse.io/v1/boards/{token}/jobs?content=true
2. Lever — api.lever.co/v0/postings/{token}
3. Ashby — api.ashbyhq.com/posting-api/job-board/{token} (token is case-sensitive — Ramp, not ramp)

For each active job posting the Actor pulls the title, location, department, full description, and public URL — then runs a case-insensitive word-boundary regex against a curated vocabulary of ~110 canonical tech names. The result is a single detected_techs column you can pivot or filter without any per-company string normalisation.

🔥 Features

• Three ATSs, one schema — Greenhouse, Lever, and Ashby normalised to an identical row shape.
• Curated tech vocabulary — ~110 canonical names spanning languages (Python, Go, Rust), frameworks (Django, React, FastAPI), databases (Postgres, MongoDB, Redis, Snowflake), cloud (AWS, GCP, Azure, Vercel), infra (Kubernetes, Terraform, Docker), CI/CD, observability, and ML/data tools.
• Per-company fault isolation — one bad token (404 from Lever, typo on Ashby) does not abort the run; the other companies still produce data.
• Pydantic v2 validation — every input and every output row is model-validated before landing in the dataset.
• Filter knobs — maxJobsPerCompany cap and minTechsDetected floor let you drop generic non-engineering postings.
• Exponential backoff — 408 / 429 / 503 retries with Retry-After honoured; up to 5 attempts per request.
• Deterministic detection — regex + vocabulary, not LLM. High precision, zero hallucinated tools.

💡 Use cases

• B2B sales qualification — enrich every Salesforce or HubSpot account with the company's live back-end stack: "they hire Django + Postgres + AWS" tells you whether to pitch your Postgres-tuning SaaS.
• Recruiter sourcing — pull every senior backend role from your target accounts and filter by detected_techs to find teams that match your candidate's stack.
• Competitive intelligence — track which competitors are hiring for Kubernetes or Snowflake and infer their roadmap before it's announced.
• CRM enrichment — replace TheirStack / BuiltWith / Wappalyzer subscriptions for the segment of buyers who publish their stack in job descriptions anyway.
• Investment research — map private-company tech footprint over time without a Crunchbase Pro seat.
• Open-source stack maps — feed a public page that ranks the most-hired-for technologies among YC-backed startups.

⚙️ How to use it

1. Open the Actor input form on the Apify Store page.
2. Add one or more {companyToken, atsType} pairs under Companies to scrape (the form prefills with airtable on Greenhouse and Ramp on Ashby).
3. (Optional) Set Max jobs per company — leave empty for no cap.
4. (Optional) Set Minimum detected techs — 2 or 3 drops generic non-engineering postings.
5. Configure Apify Proxy — leave the default residential proxy group on; we rotate sessions automatically if the ATS endpoint rate-limits.
6. Click Start. Rows stream into the default dataset as they are produced.

Quick examples

Two engineering-heavy companies, default settings:

{
  "companies": [
    { "companyToken": "airtable", "atsType": "greenhouse" },
    { "companyToken": "Ramp", "atsType": "ashby" }
  ]
}

Single Lever company, drop sales/marketing roles:

{
  "companies": [
    { "companyToken": "palantir", "atsType": "lever" }
  ],
  "minTechsDetected": 3
}

Three companies across all three ATSs, capped at 50 jobs each:

{
  "companies": [
    { "companyToken": "airtable", "atsType": "greenhouse" },
    { "companyToken": "palantir", "atsType": "lever" },
    { "companyToken": "Ramp", "atsType": "ashby" }
  ],
  "maxJobsPerCompany": 50,
  "minTechsDetected": 2
}

📥 Input

Parameter	Type	Required	Default	Description
companies	array	Yes	—	List of {companyToken, atsType} objects. atsType must be greenhouse, lever, or ashby.
maxJobsPerCompany	integer	No	unlimited	Hard cap on jobs fetched per company.
minTechsDetected	integer	No	0	Skip rows where detected_techs has fewer items than this value.
proxyConfiguration	object	No	Apify Proxy	Proxy settings. Residential group recommended for rate-limited targets.

How to find a company's ATS token:

• Greenhouse: the path segment in boards.greenhouse.io/{token} or {token}.applytojob.com. Examples: airtable, figma, stripe.
• Lever: the path segment in jobs.lever.co/{token}. Examples: palantir, netflix, eventbrite.
• Ashby: the path segment in jobs.ashbyhq.com/{Token}. Case-sensitive — Ramp, PostHog, Linear work; lowercase variants return zero jobs.

📤 Output

One row per active job posting. Rows are pushed to the default Apify dataset as they are produced — long runs surface results immediately.

Field	Type	Description
ats	string	greenhouse, lever, or ashby
company_token	string	Board slug passed to the ATS
job_id	string	ATS-canonical job identifier
title	string	Job posting title
location	string | null	Location string (Remote, NYC, Berlin…)
department	string | null	Department or team
url	string	Public job-post URL
description_text	string	Plain-text description (HTML stripped)
detected_techs	string[]	Sorted, deduplicated canonical tech names
posted_at	string | null	ISO 8601 UTC publication timestamp
scraped_at	string	ISO 8601 UTC row-creation timestamp

Sample output row:

{
  "ats": "greenhouse",
  "company_token": "airtable",
  "job_id": "4812345",
  "title": "Senior Backend Engineer",
  "location": "Remote",
  "department": "Engineering",
  "url": "https://boards.greenhouse.io/airtable/jobs/4812345",
  "description_text": "We use Python, Django, PostgreSQL, and AWS...",
  "detected_techs": ["AWS", "Django", "Postgres", "Python"],
  "posted_at": "2026-05-01T09:00:00Z",
  "scraped_at": "2026-06-01T12:00:00Z"
}

💰 Pricing

Event	Price	When it fires
actor-start	$0.05	Once per run (covers warm-up)
result-row	$0.005	Per job row emitted

A typical 50-job run costs $0.05 + 50 × $0.005 = $0.30 on pay-per-result pricing — no monthly subscription, and the back-end stack is read straight from each job description.

You pay only for rows that land. If every token is misspelled and no jobs are returned, you pay only the $0.05 start fee.

🚧 Limitations

• Tech detection is regex-based, not LLM-based — it is deterministic and high-precision, but will miss tools not in the curated ~110-name vocabulary. Submit a feedback request to add more terms.
• ~85% recall — job descriptions that omit or abbreviate tool names will not surface every dependency in a company's stack. This is an inherent limit of job-description parsing, not an Actor bug.
• Ashby tokens are case-sensitive — Ramp works; ramp returns zero jobs. Double-check the exact token from the company's Ashby board URL.
• Greenhouse double-encoding — Greenhouse wraps its content field in double HTML-encoding (<div> for <div>). The parser unescapes twice before stripping tags; unusual encoding in edge-case boards may still slip through.
• Lever descriptionPlain gaps — Lever sometimes omits the "Requirements" bullet list from descriptionPlain. The parser concatenates every lists[].content chunk to recover skills listed there; however, lists in non-standard formats may be missed.
• Default dataset retention — Apify FREE-tier default storage retains datasets for 7 days. Use Actor.open_dataset(name="…") or export immediately if you need to outlive that window.
• No personal-data extraction — this Actor reads public job posts and company-aggregated tech signals only. It does not extract candidate data or applicant details from any ATS.

❓ FAQ

Q: What makes this different from BuiltWith or Wappalyzer? A: BuiltWith and Wappalyzer sniff front-end signals — JavaScript libraries, tracking pixels, and Cloudflare headers. They are excellent for marketing-stack detection (HubSpot, Marketo, Segment) but blind to back-end infrastructure (Postgres, Kafka, Kubernetes, Snowflake). This ATS tech stack detector reads job descriptions, which is where engineering teams declare their actual data platform and server-side stack. They complement each other; they don't overlap.

Q: Is scraping these ATS job boards allowed? A: Greenhouse, Lever, and Ashby all publish their job-board APIs as official public endpoints — the same ones embedded in company career pages. We read public job post text and return company-level tech signals. No personal data, no applicant details, no authenticated endpoints.

Q: How do I find the token for a company on Greenhouse / Lever / Ashby? A: Visit the company's public jobs page. The token is the path segment: jobs.lever.co/{token}, boards.greenhouse.io/{token}, or jobs.ashbyhq.com/{Token}. Ashby tokens are case-sensitive — copy them exactly as they appear in the URL.

Q: Can I scrape hundreds of companies in one run? A: Yes. Add as many {companyToken, atsType} pairs as you need. One bad token (404 or empty board) does not abort the run — the others still produce data. Use maxJobsPerCompany to cap the volume and keep costs predictable.

Q: What happens when a company rate-limits my requests? A: We handle it. The Actor retries with exponential backoff, rotates proxy sessions on 429s and 503s, and honours Retry-After headers. If a company hits a hard block mid-run, you'll see a set_status_message with a partial count — we never silently return an empty dataset.

Q: Why does the Actor fail loud instead of returning an empty dataset? A: A silent empty result is a lie. If every token is wrong, the run exits non-zero with a clear error message — so your pipeline knows to investigate rather than assuming there were just no open roles.

🗣 Your feedback

If a tech name you care about is missing from the vocabulary, open a feedback ticket on the Actor's Apify Store page — we add new terms in batches.

Found a bug or have a feature request? Use the Store feedback form or contact DevilScrapes at https://apify.com/DevilScrapes.

---

Related Actors:

• LLM Pricing Monitor — track live API pricing across OpenAI, Anthropic, Google, Mistral, Groq, Together AI, and DeepSeek in one normalised schema.
• GitHub Org Scraper — pair tech-stack signals from job descriptions with actual code-repo activity per org.
• Y Combinator Companies Scraper — combine YC funding signal with ATS-derived tech stack for a complete startup profile.