• Implemented three algorithms—Naive Bayes, Monte Carlo Tree Search, and Q-Learning—to solve the Wordle game, leveraging probabilistic reasoning, tree-based exploration, and value iteration under an MDP framework. • Achieved a 97% win rate with the Q-Learning agent after training on 1M+ games using epsilon-greedy exploration, custom state-action encoding, and a reward function based on information gain and letter placement feedback. • Utilized Georgia Tech’s PACE HPC cluster to parallelize simulation runs and optimize training time with CUDA kernels; developed an interactive demo interface showcasing model predictions and performance metrics live with React.