This is a master’s thesis project building an intelligent Mahjong agent capable of playing against human opponents on tenhou.net. The agent uses machine learning techniques including Random Forest for decision rules (meld calls, Riichi declarations) and LSTM neural networks for waiting tile prediction. The repository also provides a framework for realtime testing and development of custom Mahjong bots, with a companion library (MahjongKit) for game log analysis and score calculation.