Overview In this role you will develop robust manipulation capabilities for robots operating in contact-rich environments, combining physics-based control with data-driven learning. You will work at scale with perception, planning, and software teams to translate research into production-ready solutions. Your work aims to enable reliable, adaptable manipulation across objects and tasks, under real-world variability. You will prototype in simulation and on hardware, and contribute to advancing robot learning and manipulation at Amazon Robotics.
Compensation / Benefits- Medical, Dental, and Vision Coverage
- Maternity and Parental Leave Options
- Paid Time Off (PTO)
- 401(k) Plan
- RSUs and sign-on rewards
- Comprehensive health and well-being programs
Responsibilities- Research, design, and evaluate ML-based manipulation policies for contact-rich tasks, integrating learning with feedback control, estimation, and motion planning
- Develop learning frameworks using simulation and real-world data to enable robust manipulation such as grasping, insertion, and object handling
- Design and run experiments in simulation and on hardware to train and stress-test policies under real-world variability
- Collaborate with software engineering to deliver scalable, real-time implementations of learning-based manipulation in production robotic systems
- Partner with cross-functional teams to transition policies from research to reliable, production-ready capabilities across Amazon Robotics platforms
Key requirements- PhD, or Master's degree and 4+ years of science, technology, engineering or related field experience
- Experience in patents or publications at top-tier peer-reviewed conferences or journals
- Experience programming in Java, C++, Python or related language
- Experience designing, running, and analyzing experiments in simulation and on real robotic hardware
- clear communication
- hands-on experimentation
- bias toward practical impact
- robot dynamics, control, and state estimation
- data-driven methods integrated with physics-based models
- reinforcement learning or imitation learning for robotics
