Senior Applied Scientist - Manipulation RL, Amazon Robotics - Vulcan Stow
Core
Design and deploy reinforcement learning policies for contact-rich, non-prehensile manipulation on physical robots to handle diverse items in unstructured environments.
Role type
Senior Applied Scientist (Robotics Manipulation)
Builds
Robotic manipulation systems for Amazon's global logistics network
Domain
Robotics, Reinforcement Learning, Control Systems
Deliverable
production ML models
Required skills
Reinforcement learning, Imitation learning, Sim-to-real transfer, Robotic control, Simulation training, Hardware integration, Technical leadership, Scientific rigor
Preferred skills
Domain randomization, Reward function design, Force/torque control, Publications in top robotics/ML venues
Technologies
Java, C++, Python
Responsibilities
Set technical direction for manipulation policies, Oversee RL approaches for diverse conditions, Manage simulation-to-real execution path, Demonstrate capabilities on real robots at scale, Establish evaluation standards and data loops, Mentor scientists and engineers, Partner with control/perception/hardware teams, Represent Amazon in academia
Seniority
Senior, hands-on IC with mentorship