The career map

From $25 an hour to $400K, built only from real job posts

Every public post from Figure, 1X, Agility, Apptronik, Skild AI, Physical Intelligence, pulled on October 1, 2026: 424 postings, 412 unique roles, read one by one.

2 of 412

jobs make a PhD mandatory

160

jobs never mention a degree

$150K–$249K

median posted pay range

1 in 13

engineering posts mention ROS

The ladder

Six real jobs. 5 of them don't mention a degree at all, the sixth accepts certifications instead. It takes years, not months, but none of the rungs asks where you went to school.

Top
Helix AI engineer, robot learning
Figure · $200K–$400K
- – deployed robot learning on real robots
- – behavior cloning or reinforcement learning
- – Python and/or C++, PyTorch
- – "analyze failures, and iterate quickly"
- – no degree mentioned
The whole ladder is the same skill growing bigger: making a real robot do something, and understanding why it failed.
on the path:M5 M6
5
Staff robotics engineer, teleoperations
Agility · $190K–$296K
- – VR, motion capture, wearables, remote control
- – whole-body kinematics, motion retargeting
- – real-time rendering and XR frameworks
- – data pipelines for learning from demonstration
- – no degree mentioned
It's rung 1 with engineering on top. Someone who spent a year as an operator understands this job's problems better than almost anyone.
on the path:M2 M3 M6
4
AI researcher, robot learning
1X · $200K–$300K + equity
- – Python and/or C++, PyTorch
- – simulation: Isaac Sim, MuJoCo or equivalent
- – trained RL policies for manipulation or locomotion
- – dealt with the sim-to-real gap on hardware
- – no degree mentioned
Control and reinforcement learning, then one real project: train a policy, measure its success rate, change one thing in the scene, measure again, write up what broke.
on the path:M4 M5 M6
3
Robot operator II
Agility · $102K–$133K
- – run Python-based test scripts
- – Linux, bash, basic networking
- – log issues and outcomes in Jira
- – associate's degree or equivalent certifications
The rung where an operations job becomes a technical job. Python every day until it's boring, then linear algebra and the robot's body.
on the path:M0 M1 M2
2
Robot service technician
Figure · $35–65/hr
- – 3+ years working with complex systems
- – multimeter, scope, current probes
- – read PCB and harness schematics
- – no degree listed
Every robot you repair teaches you how robots fail, which is exactly what the higher rungs ask about. Start writing small scripts to test what you fix.
on the path:M0
1
Data collection operator
1X · $28–36/hr
- – follow structured procedures precisely
- – comfortable working in VR, active all shift
- – between 5'5" and 6'0", to fit the VR equipment
Learn why a demonstration is bad, not just how to record one. The operator who can explain why a demo teaches the robot a bad habit moves into data quality, and later into engineering.
on the path:M5

The side doors

Imitation learning is capped by the expert it copies, so in a company that trains robots from demos, the operator sets the ceiling for the model. It's not filler work.

Robot operators

Physical Intelligence: $25/hr + benefits
Figure: $26–30/hr, night shift $32/hr
1X (data collection): $28–36/hr
Skild AI (Pittsburgh): $18–20/hr, min. 3 shifts a week

Technicians

Figure, robot service technician: $35–65/hr
1X, test technician: $30–48/hr
1X, engineering technician: $30–50/hr
1X, electrical harness technician: $40–50/hr

Programs and internships

1X AI Residency: $10,000/month, 3 to 6 months
Figure interns: $40–45/hr
PI Mechatronics Intern: strong projects and GitHubs considered early

What 226 engineering posts ask for

Python first, C++ second. ROS shows up in about one post out of thirteen, far less than beginners are told.

Python → C++ → PyTorch → a simulator → ROS, if your target needs it

Python50%
C++32%
PhD (mentioned at all)17%
"real-world robots/hardware"17%
PyTorch11%
ROS7.5%
MuJoCo or Isaac7.1%
Reinforcement learning5.8%
Imitation learning5.8%

Company by company

company	roles	no degree	what stood out
Figure	101	63	Helix AI roles up to $400K. Operators on morning, afternoon and night shifts.
1X	91	38	Pay shown on every post. $10K/month AI Residency. Many technician roles at $30–50/hr.
Agility	71	20	Robot Operator II at $102K–$133K. Several roles listed as remote.
Apptronik	69	3	Degree written into almost every post. Pay shown on only 13.
Skild AI	45	16	Operators in Pittsburgh and India, not only the Bay Area.
Physical Intelligence	35	20	Operator at $25/hr with no degree listed. Data quality, data systems, robot interfaces.

Tesla is missing: its job board doesn't offer the same public access. And "no degree mentioned" doesn't guarantee a degree won't help.

A 12-month plan with no degree

Months 1–2: Python every day, Linux and the terminal, ENGR108 linear algebra
Months 3–4: CS223A: how robot arms move. MuJoCo on a normal laptop
Months 5–6: C++ basics. AA203: control, then RL
Months 7–8: SO-101 arm ($122–$230), 50 demos with LeRobot, an imitation learning policy
Month 9: Measure it: success rate, then success rate after one change. Write up the failures
Months 10–12: Apply through a side door: operator, technician, residency, internship. Keep building at night

Two habits matter more than any course: put everything on GitHub from day one, and post what you build publicly. Robotics teams are small, and the people hiring are often the people reading.