Applied Scientist - On-Robot Learning, Amazon Robotics - Vulcan Stow
Core
Deploy and evaluate learned manipulation policies on physical robots to enable contact-rich, non-prehensile behaviors in unstructured environments.
Role type
Applied Scientist (Robotics Manipulation)
Builds
Learned manipulation policies for contact-rich, non-prehensile tasks
Domain
Robotics, Machine Learning, Control Systems
Deliverable
production ML models
Required skills
Reinforcement learning, learning-based manipulation, contact-rich control, force/torque control, Java/C++/Python programming, deploying learning-based control on physical robots
Preferred skills
Tactile sensing, compliant control, complex robotic systems at scale, publications in top robotics/ML venues
Technologies
Java, C++, Python
Responsibilities
Deploy and evaluate learned manipulation policies on physical robots; Diagnose simulation-to-reality gaps; Develop tactile and contact-aware methods; Write production-quality, real-time code; Build data collection and evaluation loops; Partner with control/perception/hardware teams; Represent organization in academia
Seniority
Senior, hands-on IC