CS undergrad specialized in high-throughput backends and autonomous reinforcement learning agents.

github

Dayson Regulus  • Peerlist

Designed and implemented a reinforcement learning agent to play Pokémon FireRed using the PPO algorithm. Trained the agent using visual cues and reward signals derived from emulator RAM data such as battle outcomes, level-ups, and map exploration. Built the training environment using Gymnasium, Stable-Baselines3, and TensorFlow in WSL (Ubuntu). Integrated multi-environment training, reward management, and model checkpointing.

AI Plays Pokémon FireRed

Developed a binary classifier using logistic regression to identify sonar signals as either rocks or mines. Performed data preprocessing, feature analysis, and model evaluation using Scikit-learn and Python. Built as part of a summer AI/ML training program under professional mentorship.