Missions

Join a cutting-edge project at the intersection of neutrino research and artificial intelligence at the Laboratoire de Physique Corpusculaire (LPC) in Caen, France. We're seeking an innova-tive postdoctoral researcher to develop Large Representation Models (LRMs) for the KM3NeT neutrino experiment as part of the nuPlato project, a partnership between LPC phys-icists and GREYC computer vision experts.

As a postdoctoral researcher, you'll be at the forefront of applying AI to fundamental physics:
• Develop a LRM for KM3NeT, a next-generation underwater neutrino telescope in the Mediterranean Sea.
• Explore generalization properties of LRMs within KM3NeT's unique and evolving datasets.
• Contribute to the emerging field of AI applications in particle physics and astrophys-ics.

Your work will directly impact our ability to process and understand neutrino data and potentially extending the techniques to other type of datasets.

Activités

Your role will encompass two key areas of development:
1. Extend and automate the current AI Software Infrastructure
o Implement and optimize MLflow for experiment tracking, model registry, and lifecycle management.
o Develop custom recipes for handling neutrino detector data and LRMs.
o Create version control systems for datasets, models, and experiment configurations.
2. LRM Development and Generalisation Studies
o Design and train transformer-based architectures for processing sparse data from neutrino detectors.
o Implement progressive training strategies to handle increasing data dimensionality.
o Conduct generalization studies across various KM3NeT detector configura-tions (ORCA-X) and telescopes (ORCA/ARCA).

Additionally, you'll have the opportunity to:
• Explore synthetic data generation using diffusion models.
• Develop interpretability and robustness analysis tools for LRMs in neutrino physics with other researchers.

Compétences

The candidate is expected to have the following skills:

Essential:
• Knowledge of Monte Carlo and advanced data analysis techniques
• Good problem-solving and practical/laboratory skills
• Experience with statistical techniques used in neutrino oscillation analyses
• Practical Experience of deploying state-of-the-art machine learning techniques

Desirable:
• Ability to develop and apply new concepts
• Verbal and written communication skills
• Ability to organize own work with minimal supervision
• Ability to work as part of a team
• Advanced computer and AI skills
• High level analytical capability
• Ability to apply relevant models, techniques and methods, and develop new ones

Technical proficiency: Pytorch, Pytorch lightning, JAX, MLFlow, GEANT4, Python, ROOT, NumPy, Matplotlib, LaTeX.

Contexte de travail

The candidate will join the KM3NeT team within the GrAMM group, which comprises two associate professors, a research director, and two CNRS research scientists. The group current-ly hosts two PhD students and two postdoctoral researchers.
Our AI initiative also involves academics from the GREYC AI groups, a couple of students and a postdoctoral researcher.
LPC CAEN, with a staff of 100, is a Joint Research Unit (UMR 6534) under the supervision of three institutions: CNRS, the University of Caen Normandy (UCN), and the École Nationale Supérieure d'Ingénieurs de Caen (ENSICAEN).

The GREYC Laboratory is a research center in digital sciences, composed of 180 members divided into 7 teams. Its activities cover both fundamental aspects and the development of practical applications.
The areas of expertise of GREYC include algorithms, computer security, artificial intelligence, image and text processing, microelectronics, etc. The laboratory is particularly involved in AI research, notably developing work on generative models to accelerate scientific discovery.

Both labs are located at walking distance from each other, on Campus 2 of the University of Caen (Campus Côte de Nacre) and are part of the ENSICAEN research park (www.lpc-caen.in2p3.fr/).

Caen is in Calvados, in the Normandy region, just a two-hour train ride from Paris. Caen is a medium-sized city located 20 minutes by car from the Côte de Nacre, offering an attractive living environment with 120 km of coastline and beaches, refined cuisine, proximity to natural parks, and direct ferry connections to southern England and Ireland.

"The framework of the AISSAI Centre's call for post-doctorates funded by the CNRS Foundation thanks to a donation from Google".

Contraintes et risques

Frequent travel in France and abroad.