Reinforcement Learning

The curriculum covers the complete RL landscape: model-free methods (Q-learning, SARSA, DQN, Double DQN, Dueling DQN, Rainbow), policy gradient methods (REINFORCE, Actor-Critic, A3C, PPO, SAC, TD3), model-based RL, and multi-agent reinforcement learning.

Each module combines theoretical explanation of the algorithm with complete Python implementation using PyTorch and standard RL environments from OpenAI Gym and MuJoCo.

The implementations are designed for readability rather than performance optimization, making them suitable for learning the algorithm logic rather than production deployment.

Visual aids reward curves, policy behavior animations, Q-value heatmaps accompany each implementation to build intuition about how learning dynamics unfold during training.

The curriculum is open-source on GitHub and freely available for individual and group study. It is used in university RL courses as supplementary material, in corporate AI training programs, and by self-taught practitioners who want hands-on RL experience beyond what theoretical textbooks provide.

The 60-day structure makes the scope of the field tractable for learners who are studying alongside other commitments, providing a concrete progression through the core algorithms that form the foundation of modern RL research and applications.