Francesco Bosso

Ai Research Engineer @Leonardo

Rome, IT

MLP-Enhanced track level clutter removal from RADAR observations. 30% accuracy improvement· Adaptive range-Doppler level clutter removal with CNN. Complete automation of semi-manual process· Parameters tuning for stochastic performance maximization. GA, PSO, REINFORCE-ADAM, GP-UCB· Creation features selection framework (wrapper, embedded). From 12 to 3 features with 1% accuracy drop· 3D Computer Vision for feature matching and Homography Transform on satellite images. From SL to SSL· Research on new methods for certification of AI trustworthiness (FAIR 5, High-quality AI)· Research on GANs for NOWCASTING advancement to trigger EU alert (collab. with Civil Protection)· CNN for onboard satellite wildfires/eruptions early detection. Contribution to PyRaws ESA ϕ-Lab repo.· Reinforcement Learning / Fuzzy Logic hybrid approach for sensor performance optimization.· Prototyping of technological innovations and handling the related IP protection (e.g. patent submission)

MIT Professional Education

Reinforcement Learning

Politecnico di Milano

Master of Science - MS