Discovering novel materials with graph neural networks and generative AI models.

● Study of generative models applied to emission tomography under the supervision of Valentin de Bortoli in collaboration with CEA. ● Pytorch multi-gpu implementation of a conditional diffusion model, continuous normalizing flow and stochastic interpolant models which enable super-resolution of PET/SPECT-scans through dose-reduction.

● Theoretical design and algorithmic implementation (Python) of a train filling simulator based on a microeconomical model.● Estimation of model parameters with supervised learning methods.● Database manipulation in SQL.

● Characteristics determination of structured products meeting the needs of the clients.● Issuance of tenders to trading rooms.● Monitoring the evolution of active structured products.

Mathematical modeling and physical computer simulations of a silicon waveguide LIDAR system.