About
I am a PhD candidate co‑supervised by Stéphane Herbin (ONERA) and Faouzi Adjed (IRT SystemX). My research focuses on making multimodal foundation models (especially Vision–Language Models) reliable under tight data constraints through PEFT (LoRA/adapters), prompt learning, in‑context learning, and distillation.
Thesis
Ensuring Frugal Adaptation of Foundation Models for Computer Vision Tasks
- Robust PEFT strategies under distribution shifts
- Minimal‑data adaptation with in‑context learning
- Trust metrics: calibration, completeness, timeliness
Publications
News
Contact
Paris, France