AI Research Scientist at Paddington Robotics Limited

Join an elite team led by a Magic Pony co-founder to build physical AI for the chaotic real world. Moving beyond lab simulations, you will leverage high-quality data from actual robot deployments to develop generalisable models for manipulation and navigation, delivering immediate commercial impact while solving fundamental challenges in robotics.

• Develop and scale 3D perception and data capture systems to help robots understand and navigate unpredictable, dynamic human environments safely.
• Design and implement cutting-edge robot learning approaches, including Vision-Language-Action (VLA) models and Reinforcement Learning for complex manipulation tasks.
• Prototype, ship, and iterate on code that runs on physical hardware, bridging the gap between theoretical research and robust real-world robotic performance.

• Holds a PhD or equivalent with 5+ years of experience publishing at top-tier venues like ICRA, CoRL, NeurIPS, or CVPR.
• Demonstrates a proven track record of deploying cutting-edge AI models onto physical hardware, showing resilience against sensor noise and environmental mess.
• Possesses strong engineering fundamentals to prototype and ship code independently while maintaining a low-ego, highly collaborative approach to rapid iterative development.

• Work under Zehan Wang, co-founder of Magic Pony (acquired by Twitter for ~$150M), alongside a world-class team from Amazon AGI, InstaDeep, and top-tier universities.
• Eschew traditional lab-only research for an "iterative deployment" approach, testing your models on real hardware in human-centric environments like supermarkets to capture unique edge cases.
• Enjoy significant ownership and early-stage equity in the UK’s leading applied AI robotics team, backed by proven founders and preparing for an upcoming funding round.