Job Posting Parser

Already have job postings — from your own ATS export, a CSV, a partner feed or another scraper — and need the same fields out of every one? Give this Actor the text (plain or HTML) and get back, per posting: the pay range as min / max / currency / period plus the yearly equivalent, the skills it asks for, the seniority from the title, the employment type, whether it is remote or hybrid, and whether it sponsors a visa. Remote and visa come with the sentence they were read from, so you can check them. Nothing is scraped: the Actor only reads what you send.

$1 per 1,000 postings, plus $0.01 per run.

The first step of the northbell job data toolkit. Each step reads data you already have:

1. Job Posting Parser (this one): posting text → pay, skills, remote, visa
2. Job Title Normalizer: title → department, seniority, core title

Why not a regex for "$"

Most dollar amounts in a job posting are not pay. Raised $20M from investors, New hire bonus: $5,000, $800 billion market — a plain pattern returns all of them. The pay reader here is the one behind our LinkedIn Salary Scraper: an amount counts as pay only when it carries a pay signal of its own (a period like "per year" or "/hr", or a label like "base salary" or "pay range"). Bonuses, funding, budgets and allowances are left out. "Up to X" and "from X" are kept at the number given.

Run it again and again

• On a schedule — save your input as a task and add a Schedule: right after your job scraper's scheduled run (call it from your pipeline through the API). Send each run to Google Sheets, Slack or email with an integration on the task.
• In bulk — Send all of the day's postings in Postings in one call.

Example

Input:

{
  "postings": [
    {
      "id": "example-1",
      "title": "Senior Data Engineer",
      "location": "United States",
      "description": "We are hiring a Senior Data Engineer. You know Python, SQL, Spark and AWS. This role is fully remote within the US. The base salary range is $150,000 - $185,000 per year. New hire bonus: $5,000. We are unable to sponsor visas for this position."
    }
  ]
}

Output (a real run of this input):

{
  "type": "posting",
  "id": "example-1",
  "title": "Senior Data Engineer",
  "salaryFound": true,
  "salaryMin": 150000,
  "salaryMax": 185000,
  "currency": "USD",
  "period": "year",
  "annualMin": 150000,
  "annualMax": 185000,
  "salaryText": "$150,000 - $185,000 per year",
  "seniority": "senior",
  "employmentTypes": [],
  "workplace": "remote",
  "workplaceText": "This role is fully remote within the US.",
  "visaSponsorship": false,
  "visaText": "We are unable to sponsor visas for this position.",
  "skills": ["Python", "AWS", "SQL", "Spark"],
  "skillCount": 4
}

The $5,000 bonus is not read as pay.

What you get

Field
salaryMin, salaryMax, currency, period	the pay as written (period: hour, day, week, month or year)
annualMin, annualMax	the same pay as a yearly figure, so hourly and yearly postings compare
salaryText, salarySource	the exact text the pay was read from; description, or salary-field when you pass a salary value
skills, skillCount	skills from a dictionary of 456 (languages, frameworks, cloud, data and AI, DevOps, security, business tools, certifications, spoken languages)
seniority	intern, entry, mid, senior, lead, manager, director or executive, from the title — the same words as every northbell job scraper
employmentTypes	full-time, part-time, contract, temporary, internship — from the employmentType you pass and the title
workplace, workplaceText	remote, hybrid or onsite only when the text says so, with the sentence
visaSponsorship, visaText	true / false only when the text says so, with the sentence

Unknown is null, never a guess. A posting that says nothing about remote work gets workplace: null, not onsite. "We build remote sensing satellites" is not a remote job.

Input

postings: a list. Each item is either the posting text as a string, or an object with description (plain text or HTML) and optionally id, title, location, salary (a pay field from your source, read first) and employmentType. Pass title when you have it: seniority and some employment types come from the title. location decides which dollar a $ is (US, Canada, Australia…). Up to 10,000 postings per run.

Pricing

$1 per 1,000 postings, plus $0.01 per run. Error rows are free. Put all your postings in one run.

For AI agents

Call with {"postings": ["<posting text>"]} — a few seconds. Each row has type: posting or error. Every field that could not be read is null.

Where the postings come from

If you do not have postings yet: LinkedIn Salary Scraper (LinkedIn jobs with pay already read) · Jobs API (461 tech companies' own job boards) · Greenhouse Jobs Scraper. For pay statistics by title and place, see the LinkedIn Salary Benchmark; for the skills a role asks for most, the LinkedIn Skills Demand report.