I'm Julien, currently based in Singapore, working at the intersection of robotics, simulation, and machine learning.

Most of my personal activity lives on Codeberg.

Some of my Linux dev setup can be found in archlinux-setup, dotfiles, and nvim.

At Naval Group, I develop a multi-agent simulation platform (Gazebo, ROS2, Unity) for Reinforcement Learning research.

• LOTUSim: Open-source real-time maritime simulation platform for multi-agent human-vehicle teaming.
• LOTUSim-Xdyn: Xdyn physics integration layer for LOTUSim, providing high-fidelity hydrodynamics.

• Hybrid-EDMD: Parametric Koopman operator learning for PDEs (heat, wave, Kuramoto-Sivashinsky). Hybrid pipeline: PGD-based reduced-order data generation in MATLAB, neural-network Koopman model K(μ) trained in TensorFlow. CNRS@CREATE.
• Branching-Light-IPT: Simulation of light branching in weakly disordered random media, via both wave optics (split-step Fourier) and classical mechanics. MATLAB. IPT 2021 competition work.
• EKT-Hubbard-Dimer: Extended Koopman's Theorem applied to the Hubbard dimer in the context of photoemission. Université de Toulouse.

• MSc in Physics (Research) — National University of Singapore. Masters' thesis on fluctuation relations and foundations of thermodynamics at the Centre for Quantum Technologies (CQT), supervised by Prof. Valerio Scarani.
• Engineering Degree — ENSTA Paris / Institut Polytechnique de Paris. Major in Applied Mathematics, Minor in Physics.