Georges Pantalos

I'm interested in reinforcement learning for robotics, autonomous driving, and language model post-training.

2024 -: co-founded AdaAI, where I worked on navigation systems for goods-to-person (G2P) warehouses trained using multi-agent reinforcement learning (MARL). I focused on cooperative pathfinding and optimal bin placement, a.k.a. "slotting".
2022 - 2024: worked on slicing algorithms for industrial 3D printing at NematX.
2021: research intern at the ETH AI Center.
2018 - 2021: M.Sc. Robotics at ETH Zurich. My thesis focused on Bayesian deep learning, supervised by Andreas Krause and Jonas Rothfuss, exploring how to build inductive bias through the data generating process.
2019 - 2020: research intern at ABB Research working on active learning for semiconductor defect detection.
2015 - 2018: B.Sc. Mechanical Engineering at EPFL.
earlier: high school math competitions (finalist in AIME, prizes in 2011 and 2013).

built an AlphaZero-style model for Quoridor, implemented in JAX for full GPU execution. You'll also find me playing basketball or chess.