My research centres on the development of testable earthquake models that forecast the space-time evolution of seismic sequences. In particular, I focus on improving the performance of physical models based on stress transfer between faults, as well as statistical forecasts based on purely empirical relationships obtained by observing how past earthquake sequences behaved. Evaluating forecast models is important to provide potentially more reliable tools for informed decision making following damaging earthquakes (e.g., for civil protection purposes). Furthermore, it serves as a testing ground for different hypotheses on which mechanisms drive earthquake triggering and cascading. I apply my models to case studies of both natural (e.g., Central Italy, Southern California) and induced seismicity and quantify their predictive skills as part of the Collaboratory for the Study of Earthquake Predictability.