Robot Learning Engineer, Manipulation (Visa Sponsorship - London, UK)

A well-funded humanoid robotics company with robots already deployed in industrial settings is hiring several senior and staff individual contributors for its Manipulation team. The team works on whole-body locomotion and manipulation using imitation learning and reinforcement learning on real hardware. The focus is practical: getting policies to work reliably on physical robots, with results shared as blog posts rather than academic papers. There is no management track.

What you'll do

• Post-train manipulation policies with behaviour cloning and RL, owning the path from raw data through to deployment on a real robot.
• Set up RL training on a digital twin with the simulation team, and iterate on reward design and simulation fidelity to achieve reliable sim-to-real transfer.
• Work with the data collection organisation to define what good data looks like, improve coverage and diversity, and refine operator instructions.
• Design data preprocessing approaches that raise the quality of collected datasets.
• Extend observation and action spaces for new capabilities, in partnership with the teleoperation and controls teams, including motion smoothness and the operator experience.
• Bring advances from the pre-training team into dependable real-world industrial demos, and feed hardware limitations back into future design cycles.

What you need

• 3+ years of robotics experience in industry or research, with shipped work on real systems (a track record, not only theory). Strong PhD candidates close to completion and postdocs are also considered.
• Hands-on experience with modern manipulation methods, ideally close to production deployment, plus a solid grasp of teleoperation and low-level robot control stacks.
• Experience with behaviour cloning and RL applied to physical systems, and familiarity with deep learning infrastructure (PyTorch or JAX, distributed training, streaming datasets, checkpointing).
• Strong software engineering fundamentals and clear communication of experiments and trade-offs. The process includes a LeetCode-style coding round.
• Proactive and debugging-focused, able to drive open problems without a detailed spec. Working English required.
• Nice to have: experience training VLA models (autoregressive, diffusion or flow-matching), RL on robotics problems, or top-tier publications or open-source contributions.

How to apply

Sign up free on i'llTip to see the employer's details and apply. i'llTip's AI can tailor your résumé to this job, and you'll get alerts for similar roles.