Deep learning · optimization · efficient training · trustworthy AI
PhD researcher @ IRIT / DEEL × SNCF
I am currently a PhD student at IRIT, within the DEEL project. I work on optimization and trustworthy AI for deep neural networks, with an emphasis on turning research ideas into fast, usable implementations.
Most of my work sits somewhere between:
optimization · privacy · robustness · distributed training
And I am familiar with:
PyTorch · JAX · TensorFlow · Triton · Linux · Slurm
deel-ai-papers/schatten-muon
PyTorch implementation of adaptive Schatten-norm optimization.
flash-newton-schulz
Fast Newton–Schulz orthogonalization with Triton kernels.
jaxlip
JAX tooling for Lipschitz-constrained neural networks.
deel-ai/lipdp
TensorFlow differentially private training with Lipschitz neural networks.
google-deepmind/optax#1602
Turbo-Muon / pre-conditioned Newton-Schulz implementation merged into Optax.
Open to Research Scientist roles — July 2027 or earlier


