About me

At Inria Paris, I'm part of the Soda team, contributing to open-source libraries like scikit-learn, skrub, and hazardous. My work on scikit-learn has resulted in a significant 20x speedup for algorithms, thanks to my focus on optimizing pairwise distance computation for sparse matrices.

I'm also proud to be part of the skrub initiative, led by Gaël Varoquaux. Our mission is to enhance data preprocessing with machine learning to optimize performance, encapsulated in our motto: “less data wrangling, more machine learning.”

Together with Olivier Grisel, we created hazardous, aiming to make survival analysis more accessible to the Python community. Our innovations include advanced estimators and tools, with notable mentions being the modified version of scikit-learn’s Histogram Gradient Boosting Tree and the SurvTRACE Transformer in PyTorch.

I've had the honor of presenting at JupyterCon 2023, EuroScipy 2023, and the Axa Dev Summit 2023, sharing insights on survival analysis and introducing skrub to a diverse audience.

My passion for knowledge-sharing has led me to teach Python for Data Science at Polytechnique and survival analysis at Dauphine University. I leverage platforms like Kaggle and Jupyterlite to ensure an interactive learning experience.

Before joining Inria, I was at the helm of ML initiatives at Spendesk, a leading European startup. My projects ranged from revenue forecasting using ensemble estimators to designing banking APIs for transaction predictions. I also co-founded Chasseur Particulier, a prominent flat-hunter agency in Paris, where I developed the tech infrastructure and implemented advanced data techniques.

I hold an MSc in Engineering from Grenoble Alpes University, INP Phelma School of Engineering, specializing in Statistics, Computer Science, and Machine Learning.

Feel free to engage with me on LinkedIn, Github, and Kaggle.