We develop a reinforcement learning framework for analog circuit design.