Short Bio

Maximilian Puelma Touzel is a Staff Research Scientist in the Applied Machine Learning Research Team at Mila – Québec AI Institute and a member of the Complex Data Lab. Puelma Touzel holds a PhD in Physics from the University of Göttingen (IMPRS for Physics of Biological and Complex Systems, Max Planck Institute), and an MSc in Physics and a BSc in Physics and Mathematics from the University of Toronto. Following postdoctoral research at the École Normale Supérieure in Paris and at Mila/Université de Montréal, Puelma Touzel has applied the same theoretical and statistical toolkit to populations of neurons, immune cells, people and, most recently, AI agents. Current research combines statistical-physics theory, statistical inference on large-scale data, and multi-agent LLM simulation to study how collective behaviour emerges in human–AI societies, with applications to information ecosystems and multi-agent AI safety. Puelma Touzel founded the STAMINA working group and co-organizes the Workshop on Social Simulation with LLMs at COLM.

Long Bio

I trained in physics and mathematics at the University of Toronto, where I worked on quantum information with A. Steinberg. Drawn to physics applied to biology, I completed my PhD with F. Wolf at the Max Planck Institute for Dynamics and Self-Organization in Göttingen, developing a statistical mechanics of neural circuit dynamics.

As a postdoctoral researcher with A. Walczak and T. Mora at the École Normale Supérieure in Paris, I built Bayesian methods that detect real change in large, noisy datasets, applied them to immune repertoire sequencing, and released them as open-source software.

I then joined Mila and Université de Montréal as an IVADO postdoctoral fellow with Y. Bengio and G. Lajoie, working on learning dynamics and reinforcement learning, including work with P. Cisek on how animals trade off reward against time. With I. Rish's Canada Excellence Research Chair in Autonomous AI, I studied multi-agent and continual learning.

Motivated by climate change and other collective-action problems, I have since focused on social systems. With the Complex Data Lab, I apply text-based inference and multi-agent modelling to public opinion, polarization and online manipulation, and I lead a team building LLM-based social simulations.

As AI agents increasingly shape how people share information and coordinate, I see multi-agent AI safety as a central question. I bring to it a combination of theory and empirical work on the collective behaviour of learning agents.

Download CV.