Data Processing/Artificial Intelligence Engineer and Applications in Particle Physics H/F

CEA; Saclay; Sur place; EUR 70 000 - 95 000; Plein temps

Résumé du poste

Irfu, based at Saclay, France, is recruiting an engineer specialized in data processing and AI/ML for its foundation model project on particle physics. You will work with physicists, computing engineers and ML researchers on developing AI methods for event reconstruction and physics analysis.

The role emphasizes self-supervised learning, HPC environments, and collaboration on publications at international conferences. Fluency in English and French is expected.

Qualifications

• PhD in computational physics, signal processing/AI or related field.
• Good knowledge of particle physics and/or nuclear physics.
• Strong background in large-scale data processing and ML/DL (neural networks, transformers).
• Proficiency with Python AI/ML frameworks (PyTorch) and scientific Python (NumPy, Pandas, Scikit‑learn).
• Experience with Git and CI/CD.

Responsabilités

• Design, train and evaluate AI model architectures for particle physics data.
• Develop self-supervised learning strategies using simulated and real LHC datasets.
• Develop data processing pipelines, experiment tracking tools and physics performance benchmarks.
• Contribute to scientific and technical documentation, publications and presentations at international conferences.

Connaissances

Python (PyTorch)

HPC / SLURM

Git / CI/CD

Transformer models

English / French comms

Formation

PhD in computational physics or AI

Outils

ROOT

HDF5

Description du poste

Irfu is recruiting an engineer specialized in data processing and artificial intelligence (AI) / deep learning as part of the development of a foundation model dedicated to particle physics to address the scientific and technological challenges of the High‑Luminosity Large Hadron Collider (HL‑LHC).

Scheduled to begin operation around 2029, the HL‑LHC will increase the integrated luminosity of the current LHC by a factor of ten, producing unprecedented volumes of data (on the order of several exabytes over its lifetime). This major evolution makes the development of new AI and algorithmic approaches essential, particularly for event reconstruction and physics analysis.

Within this context, Irfu is developing a foundation model trained on large volumes of simulated and/or real data from the ATLAS, CMS and LHCb experiments to enable efficient transfer learning (fine‑tuning) for a wide range of downstream physics tasks. This approach is directly inspired by the success of large language models (LLMs) and vision foundation models.

You will join a multidisciplinary team composed of particle physicists, scientific computing engineers and machine learning researchers.

Responsibilities

• Design, train and evaluate AI model architectures adapted to particle physics data.
• Develop self‑supervised learning strategies using simulated and real LHC datasets.
• Develop data processing pipelines, experiment tracking tools and physics performance benchmarks.
• Contribute to scientific and technical documentation, publications and presentations at international conferences.

Candidate Profile

• PhD in computational physics, signal processing/AI or a related field.
• Good knowledge of particle physics and/or nuclear physics.
• Strong background in large‑scale data processing and machine learning/deep learning (neural networks, transformers, etc.).
• Proficiency with Python AI/ML frameworks (PyTorch) and the scientific Python ecosystem (NumPy, Pandas, Scikit‑learn, etc.).
• Experience with collaborative software development tools such as Git and CI/CD.
• Experience working in high‑performance computing environments (HPC, GPU clusters, SLURM).
• Fluent in English and French, including technical English communication.

Additional assets

• Knowledge of HEP data formats and tools such as ROOT and HDF5.
• Experience with particle‑physics‑specific architectures (ParticleNet, PELICAN, Particle Transformer, JetGPT, OmniJet, etc.).
• Experience publishing scientific papers or contributing to HEP‑ML open‑source software.

Personal skills

• Ability to work effectively in multidisciplinary teams and proactively propose new ideas.
• Pragmatic, rigorous and able to tackle complex problems under multiple constraints.
• Curious, adaptable, autonomous and possessing excellent interpersonal skills.
• Excellent written and spoken communication skills in both French and English, including technical English.

Location

Saclay, Gif‑sur‑Yvette, Essonne (91), France – Ile‑de‑France, France