Rapidly growing renewable generations and peak loads pose a serious threat to the security and reliability of modern power and energy systems. A critical question is “how to accommodate a high penetration of renewable generation and deep electrification?”. This talk focuses on two key challenges, i.e., unknown information and the scalability issue of coordinating large-scale distributed energy resources. I will present two examples of developing distributed algorithms and learning-assisted control methods to address these challenges. One example is distributed model-free optimal voltage control to handle unknown physical system models, and the other is online learning and human-in-the-loop decision-making for residential demand response to deal with unknown human user behaviors.