Abstract of the talk// The growing demand for sustainable solutions to societal challenges has driven significant research efforts to integrate machine learning (ML) techniques into computational physics and chemistry. As ML becomes more prevalent in interdisciplinary research, the amount of comprehensive quantum-mechanical (QM) property data generated in recent years to train reliable predictive models has significantly increased. Recently, Leonardo and colleagues introduced high-fidelity electronic-structure data for both small and large drug-like molecules in equilibrium and non-equilibrium states. These datasets have been instrumental in advancing QM-based ML interatomic potentials (e.g., SO3LR model) and improving semi-empirical methods (e.g., EquiDTB model, thereby enabling accurate and efficient (bio)molecular simulations. In particular, recent efforts have focused on parameterizing transferable ML interatomic potentials with QM accuracy to explore the complex conformational space of RNA systems. Beyond these advances, the availability of QM structural and property data has also been key to developing novel molecular representations that enhance the accuracy and interpretability of ML models for predicting biological properties such as toxicity and lipophilicity of large drug-like molecules. In his talk, Leonardo will discuss recent developments in these areas.