Work at Dimensional

Member of Technical Staff · June 2026–present · San Francisco

At Dimensional, I work on robot learning and the infrastructure to collect data, run policies and integrate them with real robots. This work is part of dimOS, our open-source Python platform for robot applications.

• Imitation learning. I’m developing the workflow from demonstration collection and LeRobot dataset export to training tools and policy rollout. The policy runtime is integrated; the broader workflow is in progress.
• Robotics middleware. I extend configurable I/O, isolated Python runtimes and RPC tools, alongside experimental Rust SQLite/MCAP recording for reusable robot data.
• Teleoperation. I built dual-arm WebXR control with shared inverse kinematics and motion limits, connecting XR input to robot execution.
• Motion planning and execution. I develop coordinated robot motion with inverse kinematics, collision-aware planning, Cartesian paths and trajectory timing, connecting planning groups to a shared robot model.