PhD researcher working on the geometric foundations of deep learning, with applications in technical AI alignment, LLMs, reasoning models, continual learning/plasticity, and deep reinforcement learning. My work draws on ideas from statistical physics, neuroscience, and dynamical systems to better understand how intelligent systems can safely learn, reason, and retain their capacity to adapt.
Alongside my technical research, I speak about the (existential) risks associated with advanced AI, often connecting AI to broader issues such as geopolitical instability, energy, and societal resilience. I have been invited to speak at the Flemish Parliament, VRT (public broadcast of Flanders), universities, industry, etc., and I am happy to give talks on safety and AI in general to help raise awareness and improve public AI literacy.
Other interests include effective altruism, AI for science, game theory, cosmology, the evolution of cooperation, nuclear fusion, climate change, and mindfulness.