This contribution provides an overview of recent work carried out within the Data, Vibration and Uncertainty Group (https://sites.google.com/view/dvugroup) focusing on developing Physics-Enhanced Machine Learning strategies in applied mechanics. In particular, it will focus on the problem of the identification of non-smooth nonlinearities caused by frictional contacts in dynamical systems. These problems are particularly challenging because of the presence of stick-slip phenomena, the access to limited, noisy and sparse indirect measurements, and the presence of various types of uncertainty. Both numerical and experimental case studies are going to be presented in which physics and domain knowledge are integrated with machine learning architectures. Open challenges and opportunities for tackling challenges in nonlinear dynamical systems will be discussed.