Sr. Robot Learning Engineer — Planning & Manipulation

Anyware Robotics•Fremont, CA

About The Position

We are seeking a Sr. Robot Learning Engineer specializing in Planning & Manipulation to improve learned policies for robot manipulation, augmenting and extending our production system's performance envelope. This role involves contributing to our manipulation capabilities through data collection from deployed robots, physics simulation environments, policy training, and deployment validation. You will work with production multimodal data, including rich sensor streams from real industrial operations across logistics and manufacturing. Collaboration with perception and controls engineers is key to closing the loop between scene understanding, navigation, and other capabilities within AnywareOS. The models you develop will be shipped to production and run on robots at customer sites across multiple industries.

Requirements

MS/PhD in ML, robotics, or related field (or equivalent industry experience shipping learned robotic behaviors)
Strong applied ML fundamentals: policy learning (imitation learning, RL, or diffusion policies), neural network architecture design, training infrastructure
Experience with robot learning for manipulation or motion: grasp synthesis, motion generation, trajectory optimization with learned components, or sim-to-real transfer
Proficiency in Python and ML frameworks; comfortable with C++ for deployment-critical paths
Understanding of classical planning (rule-based or optimization-based planning, task-space control)
Demonstrated ability to go from research idea → working system

Nice To Haves

Prior work on force/compliance control or contact-rich manipulation
Familiarity with ROS2 and real robot deployment pipelines
Experience with sim-to-real transfer at scale (domain randomization, system identification)
Background in warehouse/logistics robotics or unstructured environment manipulation
Publications at RSS, CoRL, ICRA, or NeurIPS robotics workshops

Responsibilities

Improve learned policies for robot manipulation that augment and extend our production system's performance envelope.
Contribute to our manipulation capabilities: data collection from deployed robots, physics simulation environments, policy training, and deployment validation.
Use production multimodal data — rich sensor streams from real industrial operations across logistics and manufacturing.
Collaborate with perception and controls engineers to close the loop between scene understanding, navigation, and other capabilities inside AnywareOS.
Ship models to production — your work will run on robots at customer sites across multiple industries, not just on a sim bench.