Using generative AI to simulate chemically disordered nuclear materials at the atomic level H/F at ADSN Career

CEA is hiring a postdoctoral researcher to improve PULSE, a generative variational autoencoder method for predicting the properties of chemically disordered nuclear materials. The 24-month project focuses on increasing model accuracy for systems with thousands of atoms, adding uncertainty quantification, and generalizing the method to a continuous latent space. The researcher will work with nuclear physics and AI teams, use Python and PyTorch, and have access to CEA supercomputers, with remote work options available.