In this section, I'm proposing the reinforcement learning for system design. We made fundamental API and primitive for devops, scheduling and live migration decisions.