This post-doctoral position is part of a collaboration between LIAD (Laboratory of Artificial Intelligence and Data Sciences, CEA Saclay), the NRX Nanostructures and X-Rays Team at CEA Grenoble, the University of Lorraine, CentraleSupélec, and the European Synchrotron (ESRF). It is jointly supervised by:

·        Aurore Lomet, Research Engineer in AI at LIAD, CEA Saclay

·        Marianne Clausel (University of Lorraine), scientific lead of the national PEPR causali-t-ai program,

·        Myriam Tami, Associate Professor in Artificial Intelligence (Machine Learning and Statistics) at CentraleSupélec,

·        Joël Eymery, Head of the Nanostructures and Synchrotron Radiation Team at CEA Grenoble,

·        Jean-Sébastien Micha, Research Engineer at CNRS/ESRF,

This Post-doctoral position focuses on the development of interpretable artificial intelligence methods for detecting and analysing anomalies in synchrotron beamline experiments.

The objective is to develop Bayesian causal models and neural networks capable of identifying relevant causal relationships between instrumental parameters and observed anomalies. The work will include:

·        building hierarchical causal graphs to account for the multi-scale structure of the experimental system,

·        detecting latent variables that may affect causal inference,

·        quantifying uncertainty in causal links,

·        integrating the resulting models into neural networks (or other machine learning models) to detect and predict anomalies or anticipate failures.

The research will be based on simulated and real data collected from beamline experiments and will require collaboration with specialists in physics, optics, and instrumentation. The position involves both methodological development and applied work, with expected contributions to scientific publications and participation in collaborative meetings with the partner institutions.

PhD in AI or statistics, machine learning or applied mathematics.

Required skills and qualities:

- Fluency with Python programming for data analysis or machine learning,
- Knowledge of statistical or probabilistic modelling techniques,
- Interest in interdisciplinary work and scientific instrumentation,
- Ability to communicate effectively in English, both written and oral,
- Willingness to collaborate with researchers from different backgrounds.