I am a machine learning scientist and software engineer with deep expertise in computer vision and biostatistics.

As a Senior Research Scientist at AI biotech Owkin, I trained and deployed models for biomarker discovery, drug positioning, and cancer diagnostics — now used daily by translational teams at major pharma companies. I open-sourced the state-of-the-art model for cell detection, segmentation, and classification on histopathology slides, and published in Nature Communications and Histopathology.

Previously, I built a large-scale optimizer for sparse generalized linear models presented at NeurIPS, delivering 100x speedups over scikit-learn. At INRIA Parietal, I designed a hyperparameter calibration method for neuroimaging inverse problems, presented at NeurIPS, now part of MNE-Python.

I am an active open-source contributor and core contributor to ggml — the C/C++ tensor library underpinning llama.cpp. I built fast, portable inference engines for text-to-speech and image segmentation models.

I attended the Recurse Center batch Spring 2024. I studied in France at HEC Paris and Ecole Polytechnique.