Current opportunities.
The opportunity
Join a cutting-edge team developing humanoid robots that amplify human potential. Work at the intersection of applied deep learning and robotics to teach robots manipulation capabilities. This role involves managing the full loop from data collection and preprocessing to deployment of trained policies, collaborating closely with simulation, teleoperation, controls, and hardware teams to improve robot performance and hardware design.
What you could work on
- Develop and deploy manipulation policies via behaviour cloning and reinforcement learning.
- Design data preprocessing strategies to enhance the quality of collected data.
- Collaborate with simulation teams to improve RL training and ensure successful real-world transfer.
- Drive data collection efforts, specifying data quality, diversity, and coverage.
- Expand observation and action spaces to support new capabilities and work with Teleoperations to expose them to operators.
- Partner with Teleoperations and Controls teams to enhance motion smoothness and teleoperation experience.
- Interface with hardware design teams to communicate manipulation findings for future hardware improvements.
Relevant experience
- 3+ years working on robots with shipped artifacts.
- Strong understanding of teleoperation and low-level control stacks.
- Experience with neural network post-training and dataset curation.
- Familiarity with deep learning infrastructure such as streaming datasets, checkpointing, distributed training, and frameworks like PyTorch or JAX.
- Ability to profile, debug numerics, and write maintainable research code.
- Good software engineering practices and clear experiment documentation.
Security vetting
Confirm any vetting requirements with your consultant. An absent badge does not mean vetting is unnecessary.
Working arrangements
London Fully onsite, Full time. Confirm employment terms and any clearance requirements with your consultant.