I would like to provide an introduction to what I believe to be a thermodynamically sound coarse-grained dynamics for molecular systems: dissipative particle dynamics with energy conservation (DPDE). In this framework, one molecule is represented by a single particle with an added variable, the internal energy. I will discuss how to parametrize this model from first-principle computations and how to simulate it. One key point is to have a good model for atomic interactions, for which I will discuss one strategy to provide better descriptors in machine learning approches, based on atomic densities. I will also present some applications of DPDE to nonequilibrium systems -- namely shock waves in molecular fluids.