His research addressed a central challenge in aerospace engineering: how to reduce the prohibitive cost of high-fidelity computational fluid dynamics simulations while maintaining accuracy in complex, nonlinear aerodynamic regimes. The thesis evaluated under which conditions machine learning models can reliably predict nonlinear aerodynamics and assessed their potential as alternatives to classical projection-based reduced-order models and full-order models.

The work explored a range of advanced approaches, including Geometric Deep Learning, Bayesian Neural Networks, Volterra series, and ARMAX models, applied to both steady and unsteady aerodynamic flows. The defence committee was chaired by Prof. Gianluigi Rozza, an international expert in these fields, who expressed a very positive opinion of the research.

Congratulations to our colleague Gabriele Immordino on this important achievement!