Physical AI Engineer Intern
Summer 2026
- Full-stack ownership of a mobile robot platform, with primary focus on developing RSL-RL policies to replace a model-based QP/IK posture controller, closing the sim-to-real gap and validating outdoors on untrained terrain.
- Benchmarked reactive, recurrent and teacher-distilled architectures under one reward and observation contract; direct RL trained the recurrent policy to deployment without the two-stage distillation pipeline, cutting a stage from the standard approach.
