This document summarizes work on using deep reinforcement learning to play the Snake game. It discusses: 1) Background on deep reinforcement learning algorithms like DQN that were applied to games like Atari 2600. 2) The author's implementation of a DQN using a global/local state vector to represent the Snake environment. 3) Experiments comparing the performance of standard DQN to optimizations like Double DQN, Prioritized Experience Replay, and Dueling Network structures.