Projects

A selection of my work in data science, mathematical research and entrepreneurship. Open a project for details.

Gradient boosted trees (XGBoost): the prediction is the sum of the outputs of successive trees

Interpretable Machine Learning for Credit Risk

Building a scoring model to predict customer default on a 30-million-row, 250-variable mixed-type dataset.

Le Carnet de Prépa - my E-commerce brand (340+ sales)

Solo, I built a brand whose orders go from payment to the customer’s door with no human step. 340+ sales in year one.

Working on the problem at the Johns Hopkins Department of Mathematics

6 Months as a Full-Time Researcher: Convergence Rates for a Nonlinear PDE

Read my research report: 6 months as a Visiting Researcher at Johns Hopkins, proving exponential convergence for the Fractional Porous Medium Equation.

Neural network approximation with uniform vs gradient-aware sampling

Neural Networks applied to High-Gradient Functions

Neural networks struggle where objective functions change fast. I showed that smarter training data helps, at no extra cost.