I am an education at Ecole Normale Supérieure, a PhD in thesis in Molecular Physico-Chemistry at University Paris-Sud Orsay, a postdoctoral contract at Schlumberger Cambridge Research (UK) and at CECAM (ENS Lyon). Start at CEA in 2000, I am an expert in atomistic and multiscale simulation methods. My research activities focus on the development using machine learning tools of atomistic and multiscale simulation methodologies to describe materials. In order to improve the accuracy and transferability of interatomic potentials in Molecular Dynamics simulations, we have developed numerical potentials (GAP, SNAP and neural networks) for different materials (superhydrides, iron). Machine learning tools are also used to classify crystalline structures as well as defects (dislocations) during simulation in order to build macroscopic plasticity laws from the microscopic behavior of matter. Phase transitions, and in particular transition paths, are studied using invertible neural networks which improve statistical sampling (problem of metastability and rare events). Modeling of phase transition kinetics becomes possible. Finally, the link between atomistic simulations and mesoscopic modeling for chemically reactive systems (explosive materials) is also studied using unsupervised machine learning techniques applied to reactive Molecular Dynamics simulations.