I'm a graduate researcher working on the boundary between machine learning systems and cognition. I like problems where models have to reason, remember, adapt under constraints, or expose something useful about their internal state.

I am completing my M.Sc. in AI/ML at PSL University and working with the MILES team at LAMSADE. My current research spans information-theoretic limits in quantized LLM finetuning, early hidden-state signals in reasoning traces, and data-efficient localization of retinal symptoms in ultra-wide-field fundus images.

Outside formal research, I build tools to make ideas concrete: agent frameworks, latent-reasoning retrieval systems, from-scratch neural nets, small Android apps, and local automation utilities. I tend to learn by turning a question into a working system, then poking at the failure modes until the abstraction gets sharper.

Experience

MILES, LAMSADE / Research Intern

Paris

Researching rate-distortion limits for reconstructing reasoning solutions from quantized LLM representations, with the broader question of what information survives aggressive adaptation and compression.

INSA Lyon, DISP Laboratory / Graduate Student Researcher

Lyon

Built a semi-supervised pipeline for localizing uveitis symptoms in ultra-wide-field fundus scans, using retinal foundation models, ROI-aware preprocessing, mask-first training, and a clinical-style review interface.

Sopra Steria / Software Engineering Intern

Lyon

Designed and implemented an NLP-based CV analysis model, then shipped full-stack features for Operation, an application used by Storengy to manage workflows around France's gas network.

Stack

Centers of interest/ RL, Recursive Self-Improvement, Information Theory, Reasoning, Robotics, Metacognition, Neuroscience, Cognitive Science, AI Safety.

Languages/ Python, C++, TypeScript, JavaScript, Lua, CUDA, SQL, Java

Technologies/ PyTorch, JAX, CUDA, Hugging Face, Docker, Git, UNIX/Linux, vLLM, Ollama, OpenCV, Scikit-learn, NumPy, Pandas, PostgreSQL, Nginx

Education

ENS-PSL / Mines Paris / Dauphine / M.Sc. AI & Data Systems

Paris

Graduate work across machine learning, data systems, optimization, and modern AI research practice, with my thesis work anchored at LAMSADE's MILES team.

UCB Lyon 1 / M.Sc. CS (M1)

Lyon

Computer science master's coursework that bridged systems, machine learning, and research methodology before the PSL AI and Data Systems track.

Oregon State University / Exchange Student

Corvallis

Computer science exchange year in the United States, supported by a mobility grant and focused on broadening the technical and academic base of my undergraduate work.

UCBL / B.Sc. Computer Science

Lyon

Undergraduate computer science foundation after CPGE, completed with honors and centered on algorithms, programming, systems, and the mathematical base for later ML work.

Elsewhere

ESC
Type to search...