Applied Scientist, Amazon Robotics, Compass Team
Core
Design learning-based and model-based approaches for contact-rich manipulation in unstructured environments, ensuring safe interaction between robots and diverse physical objects.
Role type
Applied Scientist (Robotics Manipulation & Safety)
Builds
Safe manipulation algorithms and policies for Amazon's mobile manipulators and robot platforms.
Domain
Robotics, Control Theory, Reinforcement Learning, Human-Robot Interaction
Deliverable
production ML models
Required skills
Contact mechanics, force control, grasp planning, Python, C++, reinforcement learning, imitation learning, physics simulation (Isaac Gym/Sim, MuJoCo, Drake), sim-to-real transfer
Preferred skills
Control barrier functions, compliant/impedance control, dexterous manipulation, foundation models, functional safety concepts
Technologies
Isaac Gym/Sim, MuJoCo, Drake, Python, C++
