Offers “CEA”

Expires soon CEA

Stage au CEA pour Master 2 ou équivalent école d'ingénieurs H/F

  • Stage
  • Gif-sur-Yvette (Essonne)
  • IT development

Job description

Vacancy details

General information


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



Position description


Mathematics, information, scientific, software



Job title

Stage au CEA pour Master 2 ou équivalent école d'ingénieurs H/F


Physics-Informed machine learning methods applied to nondestructive testing and structural health monitoring data

Contract duration (months)


Job description

The internship will take place at the CEA, List (located on the plateau of Saclay) in the Department of Imaging and Simulation for nondestructive testing and structural health monitoring with expertise in forward and inverse problems applied to electromagnetic, ultrasound, x-ray and infrared thermography testing methods.
The Machine Learning (ML) research community can be considered one of the most active scientific community. Historically focused on time series and images interpretation (e.g., classification, regression tasks, etc.) the research community caught the interests of other scientific communities e.g., engineering and physics. In particular, in the nondestructive testing/evaluation (NDT/E) and structural health monitoring (SHM) domains the attempts to applied kernel machines and deep neural network to measurements have given positive feedbacks for solving problems that are matter of interest in the industry.
Recently, ML and engineer communities are developing a new ML paradigm based on deep neural networks to tackle the solution of forward and inverse problem in complex physics systems. That is, numerical methods historically used to model complex systems e.g., partial differential equations solutions are integrated into machine learning schemas. Such kind of approaches can be framed under the name of physics-based or physics-informed neural network (PINN). Loosely speaking, PINN approaches aim to embed the knowledge of underlying physics (e.g., wave propagation phenomena, etc.) into the learning procedure through specifically designed neural network (NN) architectures. In such approaches, the definition of a specific loss function aiming at constraining the learning procedure to the physically feasible meaningful solutions is used. Such an approach opens up different scenario in the use of NN for solving forward and inversion problems, among the most important among them being to mitigate the lack explainability (i.e., black-box models) often advocated to deep NN models.
This M2 stage aims to study the application of PINNs-like approaches in the context of NDT/E and SHM problem based on the use of simulated and/or experimental data. In such a kind of engineering domains, PINNs can be applied to physics involving waves propagation (e.g., in electromagnetic, mechanics) and heat transfer problems in complex media. The M2 stage is structured into two main parts. Due to the relevant amount of scientific and technical works on the subject, a proper taxonomy of the state of the art will be done to retrieve the most promising solutions to be proposed for solving different physical problems through PINNs-like approaches. Subsequently the implementation of the selected ML scheme retained will be applied to on infrared thermography testing problems linked to industrial-like applications with the possibility to extend the approach proposed to other NDT/E and SHM methods.

Methods / Means

Numerical simulation tools/ Matlab/ Python/ GPU computing

Applicant Profile

The candidate should have a strong background in applied mathematics and physics and good knowledge in modern machine learning with emphasis on deep learning.

Position location



Job location

France, Ile-de-France, Essonne (91)



Candidate criteria


English (Fluent)

Prepared diploma

Bac+5 - Master 2

Recommended training

Physique, apprentissage automatique, traitement su signal, mathématiques appliquée


Position start date


Make every future a success.
  • Job directory
  • Business directory