I’m an Eric and Wendy Schmidt Postdoctoral Reaserch Fellow at the I-X Centre for AI in Science, Imperial College London. Previously, I obtained my PhD from the University of Cambridge.
My research consists of merging recurrent neural networks, convolutional autoencoders, Bayesian optimization, reinforcement learning and physical knowledge to predict and control the future time evolution of dynamical systems.
Background. During my studies in space engineering, I came to deeply appreciate mathematical modelling as a tool to operate on the real world. Later on, when fellow graduate students introduced me to the growing field of machine learning, its potential and versatility impressed me to the same extent. To me, machine learning offered a new approach, complementary to existing knowledge, with seemingly infinite engineering applications. This vision and passion brought me to my current research, where I develop machine learning architectures for the temporal prediction and control of turbulent (chaotic) fluid dynamics from a dynamical systems perspective.
Experience
- 2023- : Schmdit Research Fellow, Imperial College London, I-X centre for AI in Science
- Reinforcement Learning and Model Predictive Control for the control of extreme events
- 2019-2023: PhD Candidate, University of Cambridge, Department of Engineering
- Bayesian Optimization for the selection of hyperparameters
- Physics-informed Echo State Networks to infer unmeasured physical quantities
- Prediction of extreme events with Recurrent Neural Networks
- Reduced-order modeling with Convolutional Autoencoders
- Machine learning for Probabilistic data-assimilation
- 2018-2019: Research Assistant, University of Illinois at Urbana-Champaign
- Numerical algorithms for hypersonic flows simulations
- Uncertainty quantification of reaction rates with Gaussian Processes
Education
- B.S. in Aerospace Engineering, Universita’ di Pisa, 2016
- M.S. in Space Engineering, Universita’ di Pisa, 2019
- Ph.D. in Scientific Machine Learning, University of Cambridge, 2023