Seismic resilience of infrastructure has received growing attention in the past decades. Thisdissertation aims to propose a general decision support framework for the post-earthquake damage and recovery of distributed infrastructures in order to improve their resilience. A damage assessment framework is first proposed that incorporates a neural network-based pre-event assessment model, and a graph neural network-based model for dynamically updating the damage estimates in the recovery process. Next, a spatially explicit model is developed to quantify and estimate the possible recovery trajectory based on the Gaussian Process model. Finally, a deep reinforcement learning based framework is proposed to optimize the repair actions in the recovery process. The framework can provide critical insights to the recovery of distributed infrastructure and improve pre-event planning and post-earthquake recovery tasks.