- I am francesco (braico for friends)
- I study Maths & CS for AI @ Bocconi University
- I currently am head of research/projects @ Hephaestus Applied Artificial Intelligence
- I'm passionate about ML & AI, enjoy math, am a football enthusiast, love photography, read The Economist obsessively, and devour podcast
- ποΈ Car Racing with Deep RL: I implemented a Deep Q-Learning agent for the CarRacing-v3 environment, using a CNN model with experience replay, frame stacking, and preprocessing. The agent processes raw pixel data to optimize driving actions on a racetrack.
- π Wikipedia Missing Knowledge Links: I developed a missing link analysis to identify logical but absent connections between Wikipedia pages. This process involves constructing a graph of Wikipedia pages, calculating similarity scores, clustering the graph, and applying machine learning to predict potential missing links.
- π ML Portfolio Creation & Optimization: Joint project between Hephaestus and BSIC to build a portfolio of stocks from the SP500 through ML. Implemented Principal Feature Analysis (PFA) to reduce dimensionality and XGBoost to select stocks, confronted minimum variance and minimum entropy portfolio optimization.
- π Distracted Driving Detection: I implemented a hybrid architecture that combines deep learning feature extraction with hyperdimensional computing to achieve robust multiclassification distracted driving detection from images.