Louen Pottier
AI Research Engineer · Scientific ML, Physics & 3D Vision
Research engineer at the intersection of physics and deep learning. Graduate of ENS Paris-Saclay, I have been working for three years at CEA designing neural network architectures for real-time physical simulation in virtual reality. My work has led to several scientific publications and a patent filing. Along the way I have also worked on applying deep learning to a wide variety of problems: geolocation, 3D reconstruction, verification of industrial PLC programs or car crash simulation. I enjoy exploring new fields and finding unexpected connections with my past work.
Not actively looking, but always happy to discuss new opportunities (permanent positions, freelance, teaching, scientific collaborations...)
Highlights
Preprint (2026) · first author · patent filed
LaGSplat: Latent Lagrangian Gaussian Splatting
Method to build a physics-governed interactive simulation from monocular video. Makes it possible to apply forces inside a video without a single force ever being measured.
Mistral Vibe Hackathon (2026) · 3rd overall
Impostral
Multiplayer Turing test: Mistral agents infiltrate a group of humans and try to blend in. Can you unmask the AIs before they convince everyone you are one? 🏆 1st in our category and 🥉 3rd overall, winning €1,000 in Mistral credits.
Experience
CEA-LIST
Research Engineer
Design of neural network architectures for physical simulation and model reduction applied to virtual reality (physics-informed NNs, surrogate models, world models, Gaussian Splatting). One patent filed and several scientific publications. Industrial consulting: finite element simulation substitution for Renault (crash test, GNN) and Forvia (AI for design optimization).
Read my publications → Browse my projects →
ESILV
Part-time Lecturer
Teaching fluid mechanics, thermodynamics, and partial differential equations to engineering students at Bachelor and MSc level. Courses delivered in both French and English.
See my teaching →
EDF Lab
Research Intern
AI for fluid mechanics at EDF Lab Paris-Saclay: continuous-kernel convolution layers for irregular meshes, applied to the prediction of vortical flows. Results presented at the NeurIPS 2021 workshop on Machine Learning and the Physical Sciences.
Read the paper →Skills
Machine Learning & SciML
PINNs · Surrogate Models · Representation Learning · 3D/4D Gaussian Splatting
Model Deployment
PyTorch · C++ · ONNX · TorchScript · Libtorch · Docker
Physics & Simulation
Continuum Mechanics · Model Order Reduction · FEniCS · Cast3M · Ansys
Languages
French (native) · English C1 - Cambridge English Qualification
Education
CentraleSupélec
M.Sc. - Mechanical Simulation of Structures & Coupled Systems
Specialisation in computational mechanics, finite element methods, and coupled multiphysics systems, with coursework in AI for physical simulation.
ENS Paris-Saclay
Diploma in Artificial Intelligence
Selective internal programme at ENS Paris-Saclay. Second year: research internship at EDF R&D, co-supervised by Centre Borelli (ENS AI research lab), with coursework from the MVA master's programme. Work presented at a NeurIPS workshop.
ENS Paris-Saclay
M.Sc. - Mechanics of Materials and Structures
Nonlinear mechanics, damage and fracture, continuum mechanics, and numerical methods. Simultaneously followed the first year of the ENS AI diploma (selective internal programme).
ENS Paris-Saclay
B.Sc. - Mathematics, Physics & Engineering Sciences
Intensive undergraduate curriculum in mathematics, physics, and engineering sciences at ENS Paris-Saclay (SAPHIRE programme).
Lycée Blaise-Pascal
Classe Préparatoire aux Grandes Écoles
Intensive two-year preparatory programme in mathematics, physics, and engineering sciences.