datagrit › Guides › Career Site Jobs Aggregator - ATS Salary Data

How to scrape jobs from Greenhouse, Lever, Workday and 5 more ATS at once

Monitor open roles at a list of companies across 8 applicant tracking systems in one schema, with numeric salaries and a daily only-new feed.

Published 2026-10-01 · uses the Career Site Jobs Aggregator - ATS Salary Data Actor

If you run a niche job board, a hiring newsletter or a sales team that treats hiring as a buying signal, your input is a list of companies you care about. The output you want: their open roles in one table, salary as a number, and tomorrow only the new roles.

The difficulty is that those companies do not share a careers system. One uses Greenhouse, another Lever, a third Workday, a German startup Personio. Each has its own URL pattern, its own JSON (or XML) format, its own way of writing locations and its own idea of what a salary field looks like, if it has one at all.

This post walks through what that problem looks like in practice and how to collect the data through one API call.

Where the data comes from

Applicant tracking systems publish public job board feeds so employers can show open roles on their careers pages. The Career Site Jobs Aggregator Actor on Apify reads those feeds for eight of them: Greenhouse, Lever, Ashby, Workable, SmartRecruiters, Recruitee, Personio and Workday. It uses plain HTTP, no browser and no login.

Doing this yourself means solving a set of problems per ATS:

• Finding the board. You usually know a company name or a careers page URL, not the ATS slug. The Actor reads a careers page once and follows the job board it links to most often. A bare slug is tried on Greenhouse, Lever, Ashby, Workable, Recruitee, Personio and SmartRecruiters in that order, and the first board with open postings wins.
• Different schemas. Workable has a shortcode, Workday a posting path, Greenhouse joins departments, Lever and Ashby add teams. Employment types, workplace types and locations are all written differently. The Actor maps all of them to one schema (employmentType becomes full_time, part_time, contract and so on).
• Salary in very different places. Greenhouse has pay ranges by zone, Lever a salary range, Ashby compensation tiers, Recruitee a salary field. Many other postings put the range only in the description text. The Actor reads ATS pay fields first, and otherwise looks for an explicit pay range in the description, such as 184,000 USD - 287,500 USD or $28.50 - $34.00 per hour.
• False salaries in text. Descriptions are full of numbers that are not pay: deal sizes (ACV, TCV, quota), funding amounts, grants, stipends. A range is used only with a currency and either a period word or pay wording next to it (salary, pay, base, compensation, OTE). A bare $ is not read as USD for postings in Mexico, Latin America, the Philippines, Hong Kong or Taiwan.
• Workday specifics. Listings stop at 2,000 postings per career site and each posting needs a separate detail request.

How much does description parsing add? In the measurements recorded on 1 October 2026, the description supplied the salary for 22 of 27 Recruitee postings (Great Minds), 8 of 10 Lever postings (Epoch AI) and 8 of 10 Workday postings (NVIDIA), where the ATS feed had no pay field.

Step by step

1. Define the input

Companies can be given in any form, mixed in one list:

{
  "companies": [
    "https://jobs.ashbyhq.com/ramp",
    "robinhood",
    "lever:palantir",
    "https://greatminds.recruitee.com",
    "https://nvidia.wd5.myworkdayjobs.com/NVIDIAExternalCareerSite"
  ],
  "keywords": ["engineer", "developer"],
  "excludeKeywords": ["intern"],
  "locations": ["united states", "remote"],
  "remoteOnly": false,
  "onlyWithSalary": true,
  "salaryFromDescription": true,
  "minAnnualSalary": 120000,
  "salaryCurrencies": ["USD"],
  "postedWithinDays": 30,
  "onlyNewSinceLastRun": false,
  "maxItemsPerCompany": 100,
  "maxItems": 1000
}

Title keywords match words that start with the term, so engineer also matches "Engineering". maxItemsPerCompany stops one big employer from using up the whole maxItems budget. Workday needs the full career site URL; the other systems accept a slug.

2. Run it from Python

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")

run_input = {
    "companies": ["https://jobs.ashbyhq.com/ramp", "robinhood", "lever:palantir"],
    "onlyWithSalary": True,
    "salaryCurrencies": ["USD"],
    "minAnnualSalary": 120000,
    "maxItems": 500,
}

run = client.actor("datagrit/career-site-jobs-aggregator").call(run_input=run_input)
jobs = [j for j in client.dataset(run["defaultDatasetId"]).iterate_items() if j.get("found")]

for j in jobs[:10]:
    print(j["ats"], j["company"], j["title"], j["salaryAnnualMin"], j["salaryAnnualMax"], j["salarySource"])

3. Or with cURL

curl -X POST \
  "https://api.apify.com/v2/acts/datagrit~career-site-jobs-aggregator/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"companies": ["https://jobs.ashbyhq.com/ramp", "robinhood"], "onlyWithSalary": true, "maxItems": 50}'

An input with no companies at all runs a small example (Ramp on Ashby, up to 10 postings) that is not charged per result, which is handy for checking the output shape.

What you get

One row per posting, the same schema for every ATS:

There is also jobId, requisitionId, department, team, postedAt, daysSincePosted, url, applyUrl, salaryCurrencies, equityMentioned, bonusMentioned and, on request, the plain-text description.

Recipes

Daily feed for a job board or newsletter

Set onlyNewSinceLastRun: true and schedule a daily run on Apify. The Actor remembers delivered postings per job board and per filter combination in a storage on your account, so each run returns and charges only new postings. Changing a filter starts a separate feed. Connect the run to a webhook, Google Sheets or your CMS import.

Hiring signal for sales

Feed your account list (careers page URLs work) with postedWithinDays: 7 and count new roles per company. A sudden cluster of new data or security roles at an account is a timing signal for outreach.

Salary coverage and bands in pandas

import pandas as pd

df = pd.DataFrame(jobs)

coverage = df.groupby("ats").agg(postings=("jobId", "count"), with_salary=("hasSalary", "mean"))
print(coverage)

usd_year = df[(df["salaryCurrency"] == "USD") & df["salaryAnnualMax"].notna()]
print(usd_year.groupby("company")["salaryAnnualMax"].describe()[["count", "50%", "max"]])

salarySource is worth keeping in any analysis: you may want to treat ranges written in descriptions separately from ranges in structured ATS fields.

Limitations

• SmartRecruiters has no salary or description in the list the Actor reads, so hasSalary is false and equityMentioned is null there.
• Some filters cannot apply everywhere. Workday publishes no department, so the departments filter never matches Workday postings; Greenhouse publishes no employment type, so the employment type filter never matches Greenhouse postings.
• Remote detection varies by ATS. Lever, Ashby, SmartRecruiters and Recruitee publish a workplace type and Workable a remote flag; otherwise the Actor marks a role remote only when a location says so.
• Listing caps. Workday lists at most 2,000 postings per career site; very large Lever and SmartRecruiters boards are read up to 5,000 and 10,000 postings. Cuts are reported in the run status. Workday boards are also slower, because each posting needs its own request.
• No currency conversion. Use salaryCurrencies together with minAnnualSalary.
• Slug guessing can pick the wrong board when two companies share a slug on different ATS. Use ats:slug or the board URL to be explicit; the run status shows which board each slug matched.

The Actor reads only public job board feeds that ATS vendors provide for employers; it does not log in or collect candidate data.

Wrapping up

One company list in, one schema out, with salaries you can filter on: pricing is pay per result, the Actor is on Apify, and the full field reference is in the documentation.

More guides