Efficient And Scalable Reinforcement Learning