These days, I'm mostly interested in building models that simulate complex biological systems in silico. I'm currently at South Park Commons in San Francisco, exploring how to build virtual tissues from spatial and multimodal biological data.

Previously, I spent four years at Owkin working on machine learning for cancer research. I built models for biomarker discovery and computational pathology, and open-sourced a state-of-the-art model for detecting and classifying cells in histopathology.

Before that, I worked on optimization and statistical learning at INRIA, including a sparse GLM solver presented at NeurIPS that was up to 100× faster than scikit-learn.

Outside of research, I like building fast software. I've contributed to ggml, built C/C++ inference engines for speech and vision, and was once ranked #123 worldwide on Kaggle.

I'm a Recurse Center alum (S24), and studied at École Polytechnique and HEC Paris.