CV

Research and Teaching Assistant

Toulouse INP–ENSEEIHT - IRIT

Research and teaching position.

• Ongoing research expanding PhD work on backward error analysis of neural networks.
• Research findings presented in several academic seminars.
• 200 hours of teaching in Computer Science at undergraduate and master's levels.
• Courses taught include: Optimisation, Machine Learning, Linear Algebra, High Performance Computing, and C programming.

PhD Student

University of Toulouse - Atos

Developed theoretical and experimental methods to analyse and predict the sensitivity of neural networks to various perturbations, using numerical linear algebra concepts such as conditioning and backward error.

• Introduced adversarial attacks based on parameter perturbations.
• Provided theoretical analysis of rounding errors in neural networks.
• Offered guarantees of robustness and guidelines for deep learning architectures.
• Supervised by Alfredo BUTTARI (CNRS - IRIT research director), Théo MARY (CNRS - LIP6 researcher), and Serge GRATTON (Toulouse INP–ENSEEIHT Professor).

Research Intern

Artificial and Natural Intelligence Toulouse Institute (ANITI) - Airbus

Research on rounding errors and accuracy in artificial neural networks for safety-critical embedded systems.

• Studied computational robustness for embedded neural networks in safety-critical systems.
• Compared numerical errors of different neural network implementations.
• Studied libraries for assessing numerical robustness (CADNA, FLUCTUAT).
• Developed theoretical bounds on forward error propagation through neural network layers.
• Validated theoretical results experimentally.

Computer Vision Intern

Picsellia

Implementation of an image and video annotation tool with object detection and tracking.

• Implemented motion tracking in video using Robust Principal Component Analysis (RPCA).
• Analyzed RPCA limitations with camera motion and dynamic backgrounds.
• Studied and experimented with B-SSSR algorithm improvements.
• Developed self-supervised R-CNN for class learning in automatic tracking.

Research Intern

University of Florence

Performance analysis of optimisation algorithms for training artificial neural networks.

• Conducted literature review of neural network training optimisation algorithms.
• Study and implementation of an adaptive regularisation algorithm with inexact evaluations developed by the research team.
• Applied algorithm to neural network training on benchmark datasets.
• Performed comparative performance analysis against state-of-the-art methods.
• Supervised by Stefania BELLAVIA (University of Florence Professor).

PhD in Applied Mathematics and Computer Science

University of Toulouse

Backward error analysis of artificial neural networks with applications to floating-point computations and adversarial attacks.

Research Master's Degree

University of Toulouse

Performance in Software, Media and Scientific Computing.

• Machine Learning (PyTorch, TensorFlow).
• Distributed and Parallel Computing (MPI, OpenMP).
• Data Analysis (Python, Matlab).

Engineering Degree (French Grande École)

Toulouse INP–ENSEEIHT

Major in Applied Mathematics and Computer Science, specialising in High-Performance Computing and Big Data.

• Deep Learning.
• Distributed Systems.
• High-Performance Scientific Computing.
• Optimisation.
• Statistics.

Higher school preparatory classes (CPGE)

Lycée Carnot

Preparation for entrance examinations to the national Grandes Écoles d'ingénieurs. Intensive scientific training in Mathematics, Physics and Engineering Sciences.