Description
The role develops end-to-end visuomotor policies for humanoid and general-purpose robots, including grasping, manipulation, object reorientation, door opening, dual-arm operation, and basic assembly. It covers robot learning methods such as Behavior Cloning, reinforcement learning, VLA, and policy learning; real-world data collection, dataset construction, model training, evaluation, deployment, and closed-loop optimization; simulation and digital-twin training; and integration with perception, base models, planning, control, systems, and hardware teams. The position requires a master's or doctoral degree in a relevant field, experience with real-robot learning systems, strong Python/C++ and PyTorch skills, and familiarity with robot learning and embodied AI technologies.
