Artificial Life & Minds.

Senior Researcher and Head of Artificial Life at CyberLife Technology and Creature Labs, involved from Creatures 1 (C1, 1997) through Creatures 2, Creatures 3, and Docking Station.

Co-architect of the biological computational substrate that powered autonomous virtual organisms ("Norns"). Designed heterogeneous multi-lobe neural networks with operant conditioning, a 256-chemical endocrine/receptor metabolic engine, and mutable haploid digital chromosomes enabling true digital genetics, heredity, hunger, pain, learning, and open-ended reproduction.

Technical & Creative Product Lead across four broadcast series of the BAFTA-nominated BBC children's TV show. Engineered the downloadable 3D Zook Kit, allowing hundreds of thousands of children to build physics-based virtual creatures using rigid bodies, joints, spring/damper kinematics, and neural locomotion controllers.

Contestants' winning creature designs were summoned onto a physical table in the BBC television studio using real-time broadcast mixed-reality camera tracking, competing in agility, sprint, hurdle, and push tests.

Lo-fi artificial-life simulations and cellular automata designed specifically to run in real time on low-end single-board computers (Raspberry Pi), exploring how complex emergent patterns and micro-biochemistries can thrive on extreme resource constraints.

"The body is not a given: Joint Agent Policy Learning and Morphology Evolution" (ACM / GECCO 2019). Research conducted at Google DeepMind exploring continuous co-adaptation in embodied agents.

In standard reinforcement learning, agent morphologies are fixed and only neural network weights are updated. Here, both physical 3D body structures (limb lengths, joint limits, torque multipliers, and mass distributions) and deep reinforcement learning neural policies are simultaneously evolved in competitive MuJoCo physics sumo wrestling, where two agents fight to shove each other out of a circular ring.

(Note: this is virtual morphology evolution in MuJoCo physics, not the physical EvoBorg robot).

Foundational research exploring how intelligent agents develop emergent communication, scaffold collective memory, and autonomously optimize cognitive strategies: