Modern power systems are nonlinear, complex, and interconnected with numerous heterogeneous components, causing significant challenges to system stability. Control theory techniques that provide stability guarantees typically rely on a simplified model and do not capture the nonlinear behavior of the dynamics, motivating a deep-learning-based approach. However, naive deep-learning-based approaches generally suffer from the scale of dimensionality, especially in the context of large-scale power systems. Therefore, this project aims to develop a deep-learning-based controller in a decentralized fashion based on dissipativity theory, in order to ensure the global stability of the system in a scalable fashion. (Project description, SiROP)