Postdoctoral Positions in Foundation Models for Astrophysics

The CEA Paris-Saclay astrophysics department is looking to fill 2 full-time Postdoctoral Researcher positions for the development of Scientific Foundation Models for Astrophysics, funded by the Polymathic AI initiative, and hosted within the interdisciplinary Cosmostat Laboratory at CEA Paris-Saclay. The Polymathic AI team gathers computer scientists, research engineers, and domain scientists, distributed between New York, Cambridge, and Paris, with a shared goal of developing large Foundation Models for scientific applications.

The successful candidates will work closely with Dr. Francois Lanusse at CEA Paris-Saclay and within the Polymathic AI team they will contribute to the development, production, and deployment of astrophysics-focused Foundation Models. Their overarching goal will be to deliver models that are maximally useful to astronomers and can be readily integrated within scientific workflows. This involves in particular dedicated machine learning research to handle specificities of astronomical data modalities (images, spectra, time-series, etc.) and to develop effective and reliable strategies to adapt these foundation models for downstream scientific use.

The successful candidates will have access to very significant GPU resources as well as AI expertise and engineering support, all necessary resources for building large scale models. Within the Polymathic team, they will also have access to an exciting multidisciplinary environment and will be encouraged to interact and collaborate with other members focusing on foundational ML research and on different disciplines ranging from heliophysics to biology and fluid dynamics.

Candidates should hold a PhD in astrophysics, machine learning, computer science, engineering or related technical discipline. They should also have a track record of open-source software development and machine learning publications. Candidates with previous experience in large astronomical surveys (in particular the Vera C. Rubin Legacy Survey of Space and Time (LSST)) are particularly encouraged to apply.