This small project applies Q-learning to the game of Tic-Tac-Toe using a deep neural network. As Tic-Tac-Toe is a very simple and also solved game, a simple rule-based or brute force approach could be used to easily implement an agent that plays flawlessly. Therefore the goal of this project is not to create an optimal bot for the game but to try out the concepts of reinforcement learning. The implementation is not by any means optimal.