AI/ML Hiring Signals — Track Who AI Companies Are Hiring

Monitor open roles across 98 verified AI and machine-learning companies — foundation-model labs, inference providers, AI chip startups, agent tooling, vector databases, AI coding tools and applied-AI companies — and get back more than a job list: hiring signals.

Every run tells you which roles are new since last time, which closed, whether a company is expanding or contracting, and — the one that matters most — when a company posts its first-ever role in a function it has never hired for.

A research lab posting its first Account Executive is a go-to-market signal. A dev-tools company posting its first Developer Advocate is an adoption push. A startup posting its first General Counsel is preparing for enterprise deals.

None of that is visible in a flat list of job postings. This Actor surfaces it.

---

Why this instead of a generic ATS scraper

Most job scrapers ask you to supply the company slugs. That's the hard part of the job, handed back to the buyer.

This Actor ships the company universe. You pick foundation_model or ai_agents or vector_db and it already knows who's in that category and where their board lives — including auto-detecting the ATS for companies you add yourself.

	Generic ATS scraper	This Actor
Company list	You supply slugs	98 curated AI/ML companies, categorised
Unknown company	You research the ATS	Auto-detected and cached
Output	Job listings	Listings + hiring deltas & signals
Classification	Raw ATS fields	Function, seniority, region, workplace, salary band, AI tech tags
Repeat runs	Same data again	Diff against last run

---

What you get

Job records (outputMode: "jobs")

{
  "company": "Anthropic",
  "company_category": "foundation_model",
  "company_stage": "late",
  "title": "Senior Research Engineer, Interpretability",
  "function": "ml_research",
  "seniority": "senior",
  "workplace": "onsite",
  "region": "US",
  "location_raw": "San Francisco, CA",
  "salary_min": 340000,
  "salary_max": 505000,
  "salary_currency": "USD",
  "salary_midpoint": 422500,
  "tech_tags": ["jax", "pytorch"],
  "is_new": true,
  "is_first_hire_for_function": false,
  "days_open": 3,
  "posted_at": "2026-09-10T14:00:00+00:00",
  "url": "https://job-boards.greenhouse.io/anthropic/jobs/4461450009",
  "ats": "greenhouse",
  "ats_slug": "anthropic"
}

Company signals (outputMode: "signals")

{
  "record_type": "company_signal",
  "company": "Anthropic",
  "open_roles": 214,
  "new_roles_this_run": 6,
  "closed_roles_this_run": 3,
  "net_change_30d": 18,
  "hiring_trend": "expanding",
  "gtm_signal": true,
  "first_time_functions": ["devrel"],
  "new_role_titles": ["Developer Advocate, Claude Code"],
  "roles_by_function": {"ml_research": 61, "engineering": 44, "sales": 22},
  "closed_roles": [{"title": "ML Intern, Inference", "days_open": 41}]
}

---

Who uses this

• Sales & GTM teams — hiring is the cleanest public buying signal. A company opening its first DevRel or RevOps role has budget and a new motion. onlySignalCompanies: true on a daily schedule gives you a lead feed.
• Recruiters & talent partners — see which AI companies are actively expanding, in which functions, in which regions, at what comp.
• VCs & analysts — headcount velocity by category is a leading indicator. Track net_change_30d across foundation_model vs ai_agents over months.
• Job seekers & job boards — filter 98 AI companies by function, seniority, region, remote status and salary floor in one run.
• Competitive intelligence — watch a rival's roles_by_function shift and you can see their roadmap before they announce it.

---

Input

Field	Type	Default	Notes
categories	array	["all"]	foundation_model, ai_infrastructure, ai_chips, ai_devtools, ai_agents, ai_coding, ai_data, ai_search, vector_db, ai_application, ai_platform, ai_safety, ai_robotics
stages	array	["all"]	early, growth, late, public, nonprofit
outputMode	string	jobs	jobs, signals, both
functions	array	all	ml_research, ml_engineering, engineering, infrastructure, data, devrel, security, product, design, sales, marketing, recruiting, finance, legal, operations, support
seniorities	array	all	intern → executive
regions	array	all	us, uk, eu, in, apac, ca, latam, mea
workplace	string	any	remote, hybrid, onsite
titleKeywords	array	—	Match if ANY keyword appears in title or description
excludeKeywords	array	—	Drop if ANY keyword appears
onlyNewRoles	bool	false	Needs ≥1 previous run
onlySignalCompanies	bool	false	Skip unchanged boards — ideal for daily schedules
minSalary	int	—	Filters on the top of the posted band
maxCompanies	int	60	Caps the registry slice (your own extraCompanies are never dropped)
extraCompanies	array	[]	[{"name":"Acme AI"}] — ATS and slug auto-detected

Example: daily GTM lead feed

{
  "categories": ["foundation_model", "ai_agents", "ai_coding"],
  "outputMode": "signals",
  "onlySignalCompanies": true
}

Example: senior remote ML roles paying $250k+

{
  "categories": ["all"],
  "functions": ["ml_research", "ml_engineering"],
  "seniorities": ["senior", "staff", "principal"],
  "workplace": "remote",
  "minSalary": 250000
}

---

How it works

Three public, documented job-board APIs — Greenhouse, Lever and Ashby — that companies publish specifically so their listings can be embedded elsewhere. No login, no browser, no proxies, no anti-bot evasion. A full 60-company run is a few hundred JSON requests and finishes in well under a minute at 256 MB.

Unknown companies go through a slug resolver that probes candidate slugs across all three ATSes and caches the result, so the registry heals itself over time.

Cross-run state lives in the Actor's key-value store. The first run is a baseline — it deliberately reports zero new roles rather than flagging all 800 as new. Signals start on run two.

Limits, stated plainly

• Companies on Workday, SmartRecruiters or a custom careers page aren't covered. They're skipped and counted in RUN_SUMMARY.
• Salary is only as good as what the company publishes. Ashby exposes it most reliably; Greenhouse sometimes via metadata; Lever rarely.
• region and workplace are inferred from location and description text. Good, not perfect.
• Function and seniority are rule-based classifiers. Expect ~90% accuracy on standard titles and misses on creative ones ("Member of Technical Staff" is mapped; "Chief Vibes Officer" is not).

---

Development

pip install -r requirements.txt

python -m tests.test_pipeline     # 101 offline assertions on parsing/enrichment/signals
python -m tests.smoke_run         # full Actor run against mocked ATS responses
python tools/verify_registry.py   # re-verify every company's ATS + slug (needs internet)

Deploy:

npm install -g apify-cli
apify login
apify push

---

Cost

Pure JSON over HTTPS at 256 MB RAM — a 60-company run costs a fraction of one compute unit. Comfortably inside the Apify free tier for development and scheduled personal use.