My research focuses on developing machine learning-driven approaches for the atomistic modeling of molecular and solid-state systems under real physicochemical conditions. By integrating state-of-the-art quantum mechanics at the relativistic level and molecular dynamics simulations with atomistic machine learning, I aim to deliver highly predictive models for both structural and dynamic processes.
A significant part of my work involves training graph neural networks, along with Kernel and Gaussian Process regression models, to develop machine learning interatomic potentials and machine learning NMR models. These models enable long-timescale simulations, improving our ability to analyze experimental NMR data and predict new experimental outcomes for complex materials.