Robot Learning Intern

Train imitation learning and RL policies for manipulation in simulation, and deploy them on real robots.

Use these policies to test our simulation pipelines and improve them.

Improve sim2real transfer for robots and policies.

Support the team in testing policies on real hardware: run the tests, record where and why policies fail, and define evaluation metrics that predict whether a policy will succeed on the real robot and how well it transfers zero-shot.

Hands-on experience training policies with reinforcement learning or imitation learning. If you have trained a policy in simulation and tried to make it work on a real robot, you are the right fit for us.

A current student or research graduate with an engineering background.

You have high agency and high integrity, and you are willing to learn, apply, and fail quickly.

You work with minimal guidance and do everything possible to make the product work.

It does not matter if you do not have the relevant experience in any of the requirements, as long as you have a strong interest to learn and apply quickly.

Strong Python and PyTorch.

Experience with Isaac Sim, and working knowledge of ROS2 or other robotics middleware.

Experience with MuJoCo or Genesis as well as Isaac Sim.

Sim2real techniques, such as domain randomization.

Teleoperation and demonstration data collection.

Experience deploying a policy on a real robot.