Wildfires and rainfall-induced landslides are reshaping the way we think about infrastructure resilience, with road networks often at the frontline of disruption. In hillside and fire-prone regions like Los Angeles, these hazards can trigger road closures, isolate communities, and create severe challenges for evacuation and recovery. This talk presents data-driven frameworks that combine advanced sensing technologies (LiDAR, geo-tagged cameras) with cutting-edge AI approaches, including graph neural networks and generative models, to model hazard scenarios, assess vulnerabilities, and identify optimal retrofitting strategies. Case studies from Los Angeles demonstrate how these methods provide actionable insights for enhancing the adaptability and sustainability of road networks, offering a path toward more resilient infrastructure in the face of escalating wildfire and rainfall risks.