Oscar
de Francesca

Built a large language model from first principles and fine-tuned it on public medical conversational and MCQ datasets for clinical question answering.

Designed and built an optimization engine for Decathlon's supply chain logistics, improving warehouse-to-store distribution efficiency through constraint-based algorithms.

Developed probabilistic solar irradiance forecasting models that quantify prediction uncertainty, enabling more reliable grid integration of solar energy.

Designed and built a full-stack platform for managing AI research workflows — paper tracking, experiment logging, and collaborative annotation — as my BSc thesis project.