VLA Pre-training Engineer - Deep Learning
Humanoid · UK, London
Roboticsonsitemid-level
At a glance
Highlights
- Pre-train base VLA policies on diverse multi-embodiment trajectory corpus.
- Own full loop from behaviour cloning and RL to deployment on HMND-01.
- Build pipelines ingesting synthetic data and teleop logs for retraining.
What they offer
- Stock options with meaningful upside as we scale
- 30+ paid days off including annual leave, bank holidays and shutdown
- Private healthcare including virtual and in-person care
- Pension scheme with 8% total contribution on full earnings
- Free daily breakfast, catered lunch and snacks in-office
Ideal candidate
Deep learning builder with 3+ years shipping models who loves owning training end-to-end. Curious, experimental, and eager to quickly dive into humanoid robotics and push capable policies to real deployment.
Requirements
- 3+ years building deep-learning systems with shipped models or published artifacts
- Deep hands-on experience with LLMs, VLMs, or image/video generative models
- Experience with deep learning infrastructure, distributed training and dataset streaming
- Strong Python and PyTorch/JAX with profiling, debugging and maintainable code
- Familiarity with modern software engineering practices and clear communication
Lab profile
About Humanoid
Builds commercially-scalable, safe humanoid robots for real industrial environments.
What they build
- HMND-01 humanoid platform
Known for
- Biped locomotion autonomy
- Reinforcement learning for manipulation
- Robotics simulation
- Teleoperation control
Key facts
- Operates across London, Vancouver and Boston
Posted · Last seen