Reinforcement Learning Engineer - Manipulation

Here at Humanoid, we believe in a future where robots amplify human potential. That’s why we’ve set out on a mission to build the world’s most capable, commercially-scalable, and safe humanoid robots. We’re bringing that mission to life with HMND‑01 - our rapidly developed humanoid platform being deployed in real industrial environments - and we’re growing the team to take it even further.

We're hiring a Reinforcement Learning Engineer to join our Autonomy team based in London. In this role you will leverage reinforcement learning in both simulation and physical reality to build highly performant and robust manipulation policies.

• →

  Train language-vision conditioned manipulation policies via reinforcement learning (RL) in simulation and in the real world.

• →

  Construct challenging and diverse suites of manipulation tasks in simulation.

• →

  Partner with teleoperations to collect trajectories in simulation for behavior cloning.

• →

  Partner with testing and operations to establish real-world RL training pipelines.

• →

  Experiment with various ways of bringing policies trained in simulation to the real world.

• 3+ years building deep‑learning systems (industry or research) with shipped models or published artifacts to show for it.

• Hands‑on with at least one of: LLMs, VLMs, or image/video generative models — architecture, training, and inference.

• Experience solving real problems using reinforcement learning with deep neural networks in any domain.

• Strong Python + PyTorch/JAX; you can profile, debug numerics, and write maintainable research code.

• You are self-driven, pro-active, communicate efficiently, document experiments clearly and communicate trade‑offs crisply.

• Experience with simulators for robotics (Isaac Sim, MuJoCo etc.)

• Experience in RL for robotics.

• Experience building infrastructure for large-scale RL (e.g. using ray).

• Publications at ICLR/ICML/NeurIPS or equivalent open‑source contributions.

• Familiarity with OpenVLA, Physical Intelligence (π) models, or similar open VLA frameworks.