Who's really hiring in AI — in numbers.

Every day we read the public job boards of 60 AI companies and count what's actually open: who's hiring hard, what the top roles pay, and which postings won't die. One weekly email. Numbers only, no hype.

Every row ships with a company_domain and company_linkedin_url — company website domain mapped for every row — ready to drop into Clay, Apollo or n8n.

Read straight from each company's own ATS boards (Greenhouse, Lever, Ashby, Workable) every morning — before the role reaches aggregators.

692

open AI-core roles right now

So what: every one is a live, validated opening — a conversation you can start today, not a stale listing.

AI companies tracked daily

So what: these are the boards that set the AI market — if a company isn't here, it isn't moving your clients' world.

16%

of roles are remote (112 of 692)

So what: the other 84% are tied to a place — check the location field before you promise a candidate they can work from anywhere.

21%

publish a salary range (144 of 692)

So what: the 144 that publish pay are the easiest opens — you can lead with a real number instead of a guess.

Fresh — daily

Refreshed every day at 11:30 (Vietnam time). Latest verified run: Oct 20, 2026 — 692 rows, succeeded. Hiring signals go stale in days; these don't get the chance.

Clean — 0 invalid

0 invalid rows in the latest run, and schema 1.6.0 uniform across all 692 of 692 rows. Field completeness is monitored every run — a silent data drop raises an alarm, not a shrug.

Checkable — open method

Acting on wrong data in front of a client is the fastest way to lose one. So the method is published below and the feed is public — pull the raw rows and check our arithmetic yourself.

Need the raw daily feed? Pull all live rows + company domains on Apify

Salary signals

Most AI companies still hide pay. The ones that don't are setting the market — and the top of the range is now well past half a million dollars.

OpenAI

Research Scientist / Research Engineer, Foundation — highest published range in the dataset

$380k$850k

SCALE 0 — $900K

So what: one placement at this band pays like several mid-level fills. If you run searches, this is the one worth winning.

Wayve

Principal Research Scientist, Robot Foundation Models — autonomous-driving research, near the top of the market

$370k$513k

SCALE 0 — $900K

So what: top-of-market research pay outside the US mega-labs — your candidates have a second serious bidder, and so do you.

Widest bands in this snapshot

OpenAI $380k – $850k $470k wide 56 roles

Mercor $160k – $500k $340k wide 6 roles

Perplexity $220k – $485k $265k wide 6 roles

So what: the wider the band, the more flexible the budget is for the right candidate.

Only 144 of 692 roles (21%) publish a salary range at all. When a company posts pay this high in public, it's a signal in itself: they're competing for people who have options, and they know it. So what: if you do outbound, open with these 144 — a published range means the budget is approved and the conversation starts at numbers, not at "what's the comp?"

Hiring volume leaderboard

Open AI-core roles per company, Oct 20 snapshot. Two names tower over the rest — and the interesting story is in how each one hires, not just how much.

01OpenAI

80 roles 1 of 80 remote. Also posts the dataset's top band: $850k max. So what: 80 open roles = 80 warm conversations if you sell recruiting or tooling into AI labs.

02Waymo

79 roles Deep bench at every level: 21 staff, 19 senior, 18 intern. Hiring PyTorch, JAX, TensorFlow, RL. So what: hiring at every rung means a platform team being built, not backfill — pitch the multi-role roadmap, not a single req.

03Anthropic

47 roles Clear third — nobody else is close to the top two. So what: 47 roles at a frontier lab — candidates take this call, and hiring managers here are under real delivery pressure.

04Reddit

37 roles A senior-heavy remote spree: all 37 remote, 22 staff + 8 senior. Buying experience, not headcount. So what: senior + remote is the easiest placement profile to close — no relocation, no junior ramp.

05Wayve

31 roles 0 remote, 9 senior, 8 staff — and paying research salaries that rival OpenAI's (see §01). So what: OpenAI-level pay without OpenAI's applicant flood — better odds per submission.

06Mistral AI

30 roles 0 remote and 8 lead-level roles — the most top-heavy board in the top tier. So what: lead-level searches run long and pay senior fees; worth a retained conversation.

07Anduril

29 roles 1 of 29 remote, 11 senior. Defense hires onsite, full stop. So what: onsite-only shrinks the candidate pool hard — whoever owns the local pipeline owns these placements.

08Databricks

23 roles 9 staff-level roles — top-heavy by design. So what: staff-level searches run long and pay senior fees — worth a retained conversation.

09Scale AI

23 roles 7 staff-level roles out of 23. So what: a data company hiring this senior is moving upmarket — expect the org chart, and the budgets, to follow.

10Nebius

23 roles Infrastructure-heavy: 21 of 23 senior. Skills read like a GPU lab: TensorRT, distillation, CUDA. So what: senior GPU talent is the scarcest pool in the market — if you can source it, you name the terms.

Stuck Hiring Index

The roles companies can't close — and what that says about the gap between the talent they want and the talent that exists.

Warming up

The first ranking lands mid-November

A role only counts as "stuck" with real history behind it — how many days it's been open, and whether the company quietly reposted it hoping nobody noticed. Our daily history started accumulating on Oct 15, 2026, and a fair ranking needs 35 days of it. Rather than guess, we're waiting for the data. The first Stuck Hiring Index publishes when the history matures — mid-November.

Already working: our content fingerprinting caught 2 Helsing roles re-appearing under a new board (Ashby→Greenhouse migration) — the same algorithm that will power this index.

How many hours does your team burn each week chasing candidates for roles open 60+ days — or quietly deleted and reposted?

Days openEach posting is tracked daily by a stable ID, so we see its true age — not the date on the latest repost.

Repost detectionContent fingerprinting catches the same role reappearing under a fresh listing and links it to the original.

The rankingCompanies ranked by how long their hardest roles stay open. Long-lived + reposted = struggling to hire.

Subscribe above to get the inaugural Stuck Hiring Index the day it drops in mid-November.

Skills snapshot

What 692 job descriptions actually ask for. Two skills lead the list — the market has picked its stack.

PyTorchthe default framework — assumed, not asked

243 / 692 · 35%

Agentic systemsone in four postings already asks for agents

172 / 692 · 25%

Also in demand at the top of the board: JAX, TensorFlow, RL (Waymo) · TensorRT, distillation, CUDA (Nebius).

154

staff-level roles open — the single largest seniority band. So what: companies are buying experience; lead with senior candidates, junior pipelines stall.

123

senior roles open — staff + senior = 40% of all demand. So what: two in five openings pay senior-level fees; that's where recruiting time earns its keep.

112

remote roles (16%) — but the split is company-by-company, not a trend line. So what: pitch the company, not the average — Reddit is 100% remote, Waymo 0%.

Where the roles are — and who can take them

Geography and visa policy from the daily run of Oct 20, 2026 — 692 roles, schema 1.6.0. Nearly every role in the feed carries a country tag, and more than a quarter state their visa policy in the open.

01United States

486 roles · 70%

02United Kingdom

70 roles · 10%

03Canada

36 roles · 5%

04France

23 roles · 3%

05Germany

20 roles · 3%

97%

of roles carry a country tag — 672 of 692. Location is stated in the posting, not inferred.

190

roles state a visa policy at all. The other 502 are silent on it — no policy published either way.

120

of those 190 — 63% — explicitly offer visa sponsorship.

So what: if your candidate pool is US-only, you are fishing in 70% of this market. The 120 roles that explicitly sponsor visas are a wider net many competitors skip.

Methodology & data source

Where these numbers come from

Every figure on this page comes from a daily run over the public applicant-tracking boards (Greenhouse, Lever, Ashby, Workable) of 60 AI companies. Roles are filtered to AI-core positions — AI/ML engineering, data and AI product — and enriched with parsed salary ranges, seniority, workplace type and skills. Every row also carries a company website domain and company_linkedin_url, mapped for every row — so GTM teams can route these signals straight into Clay, Apollo or n8n without a cleanup pass. Each role carries a stable ID, so openings, closures and reposts are tracked day over day instead of guessed from a single scrape.

Remote means the role's workplace_type as reported by the company's own ATS (remote / hybrid / onsite). It is not inferred from the location string. 279 of 692 roles (40%) state no workplace policy, so they are counted as neither remote nor onsite — the 16% remote figure is a floor, not a ceiling.

The underlying dataset is public: the same feed powers the AI/ML Jobs Scraper on Apify, where anyone can run it, inspect a run, or pull the raw rows and check our arithmetic.

Data snapshot: Oct 20, 2026
Schema version: 1.6.0
Roles in snapshot: 692 AI-core
Companies tracked: 60

Snapshot: the daily run of Oct 20, 2026 — 692 rows, all figures on this page. Quality gate: 0 invalid rows, schema 1.6.0 across 692 of 692.