Where Remote AI Jobs Actually Live in 2026
Most “remote AI job” searches return noise. Stale listings, recycled postings, syndicated ghosts of roles that closed months ago.
We skipped the job boards. We pulled directly from Greenhouse and Lever—the applicant tracking systems where real hiring actually happens. As of late April 2026: 448 open AI/ML roles across 89 companies.
Here’s what the data shows.
Five Companies Control Nearly 40% of the Market
If you’re not looking at these five, you’re working the wrong end of the funnel.
Scale AI — 53 roles. Moved from data-labeling into AI evaluation and red-teaming. Hunting applied engineers and strategy consultants. The recruiting process feels like a filter, not a conversation.
Databricks — 52 roles. Enterprise. Competitive. The kind of company where a phone screen takes three weeks to schedule—and the compensation reflects exactly why you tolerate that. Heavy MLOps and platform engineering needs.
Reddit — 28 roles. ML at scale. Recommendation systems, trust-and-safety, the full stack of keeping 100M+ daily users from breaking everything. Serious engineering culture.
Cresta — 26 roles. Building AI copilots for customer-facing teams. Smaller and faster than the others, hiring at a pace that signals real momentum.
Applied Intuition — 19 roles. Simulation and robotics. Mountain View HQ, but hiring distributed. Niche focus, low competition for candidates.
The Companies Worth Watching Below the Radar
PointClickCare (15 roles) — ML applied to healthcare operations. Quiet. Underrated.
Coreweave (15 roles) — GPU cloud infrastructure. The pick-and-shovel play in AI. Less name recognition, fewer applicants per role.
Snorkel AI (12 roles) — Programmatic data labeling. Unglamorous. Essential to everything above it.
Figure AI (11 roles) — Humanoid robotics. Embodied intelligence. Early, high-risk, high-signal.
Airtable (8 roles) — AI layered into structured data and workflow. Stable company, growing AI surface area.
Between all ten, 239 of 299 remote-eligible roles are accounted for.
What “Remote” Actually Means in This Data
Here’s where it gets uncomfortable: 257 of 448 roles—57%—are listed as “Remote / Unknown.”
That’s not a promise. It’s usually an unoptimized default field in an ATS that nobody cleaned up. The true count of purely remote-first roles is likely 80–150 once you read the fine print.
The big five have real distributed programs. For everyone else, treat the job description like a legal document. Hybrid requirements, timezone restrictions, and country locks are buried in there. Candidates who miss that lose weeks chasing the wrong leads.
How to Work the Market
Search remote-eligible, not remote-confirmed. Filtering only for your city cuts out more than half the available roles. Cast wider, then read carefully.
Go where the volume is. Imperfect fit at Scale or Databricks beats a perfect fit at a company with one seat open. More roles means more surface area to find a way in.
Check Monday, not Friday. The index updates every Monday. By end of week, the best leads already have pipelines behind them. Browse Machine Learning Engineer and LLM Engineer roles at the start of the week.
Watch the quiet ones. Coreweave and Snorkel AI don’t have brand recognition driving thousands of applicants to every post. That’s an advantage—if you’re paying attention.
The companies at the top of this list will fill these roles in weeks. A different set will open. The market doesn’t wait.
HireIndex pulls live AI/ML hiring data from Greenhouse, Lever, and LinkedIn—updated every Monday.