General information
Organisation
The French Alternative Energies and Atomic Energy Commission (CEA) is a key player in research, development and innovation in four main areas :
• defence and security,
• nuclear energy (fission and fusion),
• technological research for industry,
• fundamental research in the physical sciences and life sciences.
Drawing on its widely acknowledged expertise, and thanks to its 16000 technicians, engineers, researchers and staff, the CEA actively participates in collaborative projects with a large number of academic and industrial partners.
The CEA is established in ten centers spread throughout France
Reference
2026-41672
Description de l'unité
The CEA is at the heart of society's challenges, particularly the energy transition. In this field, cutting-edge research is being carried out at the Institut de Recherche sur la Fusion par Confinement Magnétique (IRFM), whose aim is to develop a sustainable, environmentally-friendly energy source based on the use of fusion energy found in stars.
To produce this energy on Earth, we need to heat a medium called plasma to several hundred million degrees in high-tech facilities such as tokamaks. This challenge, rich in industrial and economic prospects, requires major scientific and technological advances that are mobilizing the scientific community. The IRFM operates the WEST tokamak as part of the European fusion program, in preparation for future experiments on the international ITER tokamak currently under construction on the Cadarache site. IRFM researchers, together with their academic partners, are also developing the theoretical and modeling tools needed to understand the phenomena at the heart of fusion plasmas, while engineers and technicians are working on innovative technologies in fields such as cryomagnetism, high-frequency wave heating and new materials for extracting intense heat fluxes.
Position description
Category
Mathematics, information, scientific, software
Contract
Postdoc
Job title
Postdoctoral Researcher in AI – Self-Supervised Learning of Multimodal Representations (M/F)
Subject
Development of self-supervised learning and multimodal representation methods for the analysis of experimental data from multiple diagnostics of the WEST tokamak, with a particular focus on multivariate time series and their time-frequency representations.
Contract duration (months)
24
Job description
Responsibilities
Within a multidisciplinary team combining AI, signal processing and plasma physics, the main responsibilities will be to:
- Develop and evaluate self-supervised learning and multimodal representation methods suited to heterogeneous multi-sensor experimental data.
- Investigate the ability of learned representations to extract relevant information about plasma behavior and dynamics from multivariate time series and time-frequency representations, particularly in situations where annotations are limited.
- Define, in collaboration with WEST physicists, several scientific use cases to assess the usefulness, robustness and generalization of the learned representations.
- Compare the developed approaches with recent state-of-the-art methods and, where relevant, explore their methodological transferability to other multi-sensor scientific datasets.
- The work will result in a documented methodological framework, reproducible demonstrators based on WEST and geophysical data, as well as scientific publications.
Work environment
The position is based at the CEA/IRFM in Cadarache and will be carried out in collaboration with the CEA/DAM/DIF in Bruyères-le-Châtel.
The postdoctoral researcher will work in an interdisciplinary environment combining artificial intelligence, signal processing, experimental data analysis and plasma physics. They will interact with the scientific and technical teams involved in the analysis of WEST data, as well as with the DAM/DIF teams.
The contract is for a period of 24 months, with a target of two scientific publications during the project.
Methods / Means
Python, PyTorch, self-supervised learning, multi-sensor data processing and Git.
Applicant Profile
PhD in machine learning, applied mathematics, signal processing, computer science, applied physics or a related field, preferably with the PhD awarded within the last two years.
You have solid experience in machine learning and/or deep learning, with strong skills in Python, PyTorch and scientific computing. Experience with multivariate time series analysis, signal processing, multi-sensor data, representation learning, self-supervised learning or multimodal learning would be particularly appreciated.
Experience in working with experimental data from large-scale scientific facilities or complex multi-sensor systems would be an asset.
Proficiency in scientific software development best practices (Git, documentation, reproducibility of experiments), a good level of scientific English, and the ability to work collaboratively in an interdisciplinary environment are required.
Autonomy, scientific rigor and a strong interest in applying modern artificial intelligence methods to physics are expected.
In accordance with the CEA's commitments to the integration of people with disabilities, this position is open to all applicants. The CEA offers appropriate accommodations and/or organizational arrangements to support the inclusion of employees with disabilities.
Position location
Site
Cadarache
Job location
France, Provence-Côte d'Azur, Bouches du Rhône (13)
Location
Saint Paul lez Durance
Candidate criteria
Languages
English (Fluent)
Recommended training
PhD in machine learning, signal processing, applied physics or a related field.
Requester
Position start date
01/12/2026
Create candidate area
