Deep Learning - Robot Manipulation Engineer

Job description

Job responsibilities

• Design and implement advanced deep learning models and training procedures to achieve dexterous manipulation for humanoid robots with high DOF and multi-fingered hands.
• Train models, using curriculum learning strategies, to progress from simple object interactions to precise grasping, long-horizon tasks, tool-use, and in-hand object manipulation.
• Incorporate tactile sensing and proprioception into end-to-end learning pipelines, enabling robust closed-loop policies.
• Work with the teleoperation and data team to design data collection and versioning strategies.
• Leverage existing state-of-the-art manipulation models and contribute to the development of new architectures for emerging complex tasks.
• Deploy trained manipulation models to robotic hardware, ensuring real-time performance, safety, and integration with control systems and sensors.
• Collaboratively develop and optimize the manipulation ML pipeline.
• Keep up to date with the state of the art in research and development.

Job requirements

• Courage and grit to tackle some of the hardest problems in robot manipulation.
• Enthusiasm for working collaboratively in fast paced ambiguous environments.
• Masters or PhD in Robotics, Computer Science, or a related field.
• 3+ years of experience in applying deep learning to robotic manipulation.
• Strong understanding and proficiency with state of the art algorithms and best-practices in behavior cloning, vision-language-action models, diffusion policies, foundation models, etc.
• Experience with cloud computing and large-scale datasets.
• Understanding of the challenges of deploying neural network models in the real world.
• Capable of writing high quality software.