This lecture covers the development of Car-Parrinello molecular dynamics, a unified scheme combining molecular dynamics and density-functional theory, allowing the simulation of covalently bonded and metallic systems. The presentation includes the historical background, technical details, and the impact of this approach on materials modeling. It also discusses the challenges in atomistic simulations, the statistical mechanics behind the method, crystal transformations, and the integration of quantum mechanics. The lecture concludes with the application of machine learning to enhance ab-initio molecular dynamics simulations.