Weekly measurement, snapshot September 10, 2026
The Physical AI Hiring Index
Who is hiring the people who build embodied AI, measured weekly from employers' own job boards.
Who is hiring the people who build embodied AI?
As of September 10, 2026, 764 open technical roles sit at 10 of the 10 tracked embodied-AI employers, out of 1,201 postings scanned at the companies building AI that acts in the physical world. Waymo (221), Zipline (168) and Nuro (86) carry the most. Functional safety standards is the most named standard (18 roles). Median disclosed pay is $161,750 (16 roles disclose it). Every posting links to the seat on our map it matches, and every seat to the graded program that closes the gap.
01 Demand
How many physical AI jobs are open right now?
764 as of September 10, 2026, counted from the public job boards of 10 tracked embodied-AI employers, 63.6 percent of the 1,201 postings scanned. 451 of them are senior, director or executive level. Only postings whose title names the work are counted, so this is a floor rather than a ceiling.
| Role family | Open roles |
|---|---|
| Perception and autonomy | 267 |
| AI and machine learning | 231 |
| Hardware and embedded | 118 |
| Safety and assurance | 68 |
| Planning and control | 48 |
| Simulation and data | 30 |
| Other | 2 |
| Seniority | Open roles |
|---|---|
| Senior or lead | 416 |
| Mid-level | 263 |
| Entry or associate | 50 |
| Director or principal | 29 |
| Executive or head | 6 |
02 Employers
Which companies are hiring the most physical AI talent?
Ranked by open technical roles this week: 221 of 336 open roles at Waymo, 168 of 349 open roles at Zipline, 86 of 110 open roles at Nuro, 68 of 83 open roles at Apptronik and 63 of 103 open roles at Figure AI. Each employer page lists its roles and the standards they name. The second figure is every open role at that employer, so you can see who is building technical capacity relative to their size.
| Employer | Technical roles | All open roles |
|---|---|---|
| Waymo | 221 | 336 |
| Zipline | 168 | 349 |
| Nuro | 86 | 110 |
| Apptronik | 68 | 83 |
| Figure AI | 63 | 103 |
| Motional | 58 | 69 |
| Agility Robotics | 42 | 67 |
| May Mobility | 35 | 45 |
| Dexterity | 18 | 31 |
| Locus Robotics | 5 | 8 |
03 Standards
Which standards and regulations show up most in physical AI job postings?
Inside the technical roles: Functional safety standards (18), GDPR (3), EU AI Act (1) and EU Machinery Regulation (1). Across every posting scanned at the tracked employers, technical role or not, the standards named most are Functional safety standards (20), GDPR (4) and UL 4600 (autonomous systems safety) (2). Each standards page lists the roles that cite it and the graded competencies that answer it.
| Standard or regulation | Technical roles naming it |
|---|---|
| Functional safety standards | 18 |
| GDPR | 3 |
| EU AI Act | 1 |
| EU Machinery Regulation | 1 |
| UL 4600 (autonomous systems safety) | 1 |
04 Skills
What skills do physical AI employers ask for?
Read from the same postings: robot learning and foundation models (109), embedded and edge AI (83), functional safety and assurance (72), perception and computer vision (66), fleet operations and teleoperation (54) and motion planning and control (53). Each maps to graded competencies on the careers map, and every posting in the feed links to the seat it matches.
- Robot learning and foundation models109
- Embedded and edge AI83
- Functional safety and assurance72
- Perception and computer vision66
- Fleet operations and teleoperation54
- Motion planning and control53
- Machine learning infrastructure39
- Simulation and sim to real31
- Human robot interaction12
In the vocabulary of our map, the competencies these postings call for most: Applied AI engineering, Embodied AI and robotics systems, AI evaluation and testing design, How models work, at a governance depth, Post-deployment monitoring and drift detection, Physical AI safety assurance.
05 Seats on the map
Which roles on the careers map are these postings for?
242 of the 764 postings carry a title the map recognizes, matched by the titles postings really use for each seat. The rest are counted in the index and shown in the feed with the generic readiness check. Each role page shows its open postings, the competencies employers name and the readiness check for that seat.
- AI Engineer, production systems edition Evaluation and engineering173 open postings
- Robotics AI Engineer, embodied systems edition Evaluation and engineering39 open postings
- AI Product Manager, responsible product edition Product, program and transformation8 open postings
- AI Program Manager Product, program and transformation6 open postings
- AI Evaluation Specialist Evaluation and engineering5 open postings
- Physical AI Safety Engineer, autonomous systems edition Risk, audit and assurance4 open postings
- Operations Manager, AI-ready Workforce and enablement3 open postings
- AI Adoption and Enablement Lead Workforce and enablement1 open posting
- AI Governance Engineer Evaluation and engineering1 open posting
- AI Security Architect Security and resilience1 open posting
- Security Compliance Manager Security and resilience1 open posting
All 50 roles on the map, including the seats no tracked employer posted this week.
06 Pay
What do physical AI jobs pay?
Among the 16 technical roles that disclose a US dollar range, the median midpoint is $161,750, with the middle half between $137,500 and $197,500. Disclosure is concentrated in US employers, so this is the top of the market, not a global average.
07 Newest
The newest physical AI postings this week
All 764 roles, filterable by standard, family, seniority, employer and remote
08 Your move
The shortage is measured, and the measured part is the governance half.
Recruiters count the engineers. What these postings also show, week after week, is employers asking for safety cases, standards evidence and AI Act readiness in the same breath as ROS 2 and reinforcement learning. That overlap is the seat our map trains for. Pick a posting, open the readiness check for its seat, and the path is merged from lessons that already exist. Nothing is authored per person, and the first module of every program is free.
- Certified AI Practitioner: Workplace Foundations
97 graded topics answer 32 of the competencies this week's postings name. First module free.
- Certified AI Governance Professional (CAIGP)
57 graded topics answer 28 of the competencies this week's postings name. First module free.
- The AI Lobbyist: Certified AI Policy Strategist
35 graded topics answer 28 of the competencies this week's postings name. First module free.
- Certified AI Transformation Professional (CATP)
49 graded topics answer 23 of the competencies this week's postings name. First module free.
Companion measurement: the AI Governance Hiring Index tracks the governance half of the same market at the AI labs and platforms.
09 Method
How the index is built
Every week an automated crawler reads the public applicant-tracking-system feeds (Greenhouse and Lever) of 10 tracked embodied-AI employers, keeps every posting whose title names engineering, research, data or safety work, and tags each against a fixed taxonomy of 10 standards and regulations and 10 skill families (version 2026-09-10.1). Sales, corporate, trades and technician titles are excluded by rule. Full descriptions are read for tagging and never republished.
Sources
The public job-board feeds of 10 embodied-AI employers: humanoid and warehouse robotics, autonomous vehicles and delivery drones. 10 answered this crawl. Nothing is scraped from LinkedIn or Indeed. Every role links to the employer's own posting.
Classification
A posting enters the index when its title names engineering, research, data or safety work; sales, corporate, trades and technician titles are excluded by rule. The body is then read for 10 standards and 10 skill families with fixed patterns, after the equal-opportunity and benefits boilerplate is dropped. Titles are matched to the careers map through each role's own title variants.
Cadence and history
Weekly. Every snapshot is archived, so the series can show standards-citation velocity and per-employer hiring intensity over time. Taxonomy changes are versioned (this snapshot: 2026-09-10.1) so a rule change is never mistaken for a market move.
Limits, stated plainly
Coverage is the tracked employers, not the market: Tesla, Boston Dynamics, 1X and Sanctuary AI expose no public board feed and are not counted, and the automotive and aerospace incumbents are not yet crawled. Pay figures are the postings that disclose a US dollar range, a small minority at these employers. Classification is deterministic, so precision is high on title-level roles and a role titled unusually can be missed.
10 Data
Download and cite
The dataset is free to reuse under CC BY 4.0. Three files, rewritten weekly:
- postings.json: every technical role with employer, link, location, standards, skills, disclosed pay and matched seat. Never the description.
- index.json: every figure on this page, plus the taxonomy that produced it.
- companies.json: the tracked employers and each one's counts.
Cite as: GAGE (Global Academy of Generative-AI Education), "Physical AI Hiring Index", snapshot 2026-09-10, https://www.gage.academy/careers/physical-ai-index. Reuse under CC BY 4.0.
Questions
- How many physical AI jobs are open right now?
- 764 as of September 10, 2026, counted from the public job boards of 10 tracked embodied-AI employers, 63.6 percent of the 1,201 postings scanned. 451 of them are senior, director or executive level. Only postings whose title names the work are counted, so this is a floor rather than a ceiling.
- Which companies are hiring the most physical AI talent?
- Ranked by open technical roles this week: 221 of 336 open roles at Waymo, 168 of 349 open roles at Zipline, 86 of 110 open roles at Nuro, 68 of 83 open roles at Apptronik and 63 of 103 open roles at Figure AI. Each employer page lists its roles and the standards they name.
- Which standards appear most in physical AI job postings?
- Inside the technical roles: Functional safety standards (18), GDPR (3), EU AI Act (1) and EU Machinery Regulation (1). Across every posting scanned at the tracked employers, technical role or not, the standards named most are Functional safety standards (20), GDPR (4) and UL 4600 (autonomous systems safety) (2). Each standards page lists the roles that cite it and the graded competencies that answer it.
- What skills do physical AI employers ask for?
- Read from the same postings: robot learning and foundation models (109), embedded and edge AI (83), functional safety and assurance (72), perception and computer vision (66), fleet operations and teleoperation (54) and motion planning and control (53). Each maps to graded competencies on the careers map, and every posting in the feed links to the seat it matches.
- What do physical AI jobs pay?
- Among the 16 technical roles that disclose a US dollar range, the median midpoint is $161,750, with the middle half between $137,500 and $197,500. Disclosure is concentrated in US employers, so this is the top of the market, not a global average.
- How is the Physical AI Hiring Index built?
- Every week an automated crawler reads the public applicant-tracking-system feeds (Greenhouse and Lever) of 10 tracked embodied-AI employers, keeps every posting whose title names engineering, research, data or safety work, and tags each against a fixed taxonomy of 10 standards and regulations and 10 skill families (version 2026-09-10.1). Sales, corporate, trades and technician titles are excluded by rule. Full descriptions are read for tagging and never republished.
- How often is it updated?
- Weekly. Each crawl refreshes every count on this page and the feed, and each snapshot is archived, so the series shows how embodied-AI hiring moves over time.
Looking for a seat that is not here this week? Where the jobs are carries live searches for every role on the map, and each role page shows what the seat asks for.