Simplify& Improve
Building systems that let machines understand the world — and recover when it breaks.
A machine learning engineer drawn to robotics — happiest when building things that work in the real world.
I'm a research assistant at Imperial College London, focusing on world-models, reinforcement learning, and damage adaptation. Before research, I led and shipped products — so I care as much about the system around the model as the model itself.
Less moving parts, sharper results. Simplify, then improve.
AWARE
Detecting anomalies in robotic systems with only CCTV cameras — an air-gapped monitoring resolution for legacy, critical infrastructure.
Selected work
Robust Obstacle-Traversal Agility for Damaged Robots
Single generalist locomotion policy with damage adaptability.
Quality-Diversity Reinforcement Learning for Damage Recovery in Robotics
Helping legged robots recover from unexpected damage with better performance and far fewer trials — ReX-MAP-Elites.
Driving Condition-based Energy Management Strategy of Hybrid Vehicles
A driving-profile-aware strategy that lowers the energy consumption of hybrid electric vehicles via multi-agent DDPG.