Large earthquakes occurring at converging tectonic plate boundaries pose significant threats to communities worldwide. These so-called subduction zone earthquakes can trigger tsunamis and cause widespread damage, but scientists still cannot predict when they will occur. The most damaging events happen rarely, often centuries apart in a region, and their observations are scarce and rely on indirect inference. While physics-based models of subduction zones help explain these phenomena, they struggle to capture how small and large earthquakes interact over a wide range of timescales. To better understand megathrust systems and assess hazards, this project will develop new computational tools that combine real-world geophysical observations and models based on physics. The research team will study major earthquakes in Chile, Japan, and Alaska that occurred over the past 20 years to investigate their immediate and long-lasting impacts in these regions. Through partnerships with local organizations and teachers, the team will develop community-oriented learning materials for university courses as well as high schools and organize outreach events. The project will support and train undergraduate interns, graduate students, and a postdoctoral scholar, contributing to training and inspiring the next-generation geoscience workforce. This project will advance state-of-the-art methods for inferring and modeling megathrust faulting processes across timescales ranging from seconds to ce