Adaptive learning systems offer a transformative approach to education by tailoring content delivery to individual learners’ needs and abilities. This study proposes a Deep Q-Learning (DQN)-based framework to address challenges such as scalability, engagement, and personalization in adaptive learning environments. By modeling the learning environment as a Markov Decision Process (MDP), the system dynamically optimizes learning paths through continuous evaluation of learner profiles, engagement metrics, and performance outcomes. Using the EdNet dataset, a large-scale educational dataset with over 131 million learner interactions, the proposed framework demonstrates significant improvements in learning outcomes, engagement metrics, and decision-making effectiveness. Key training techniques, such as experience replay, target networks, and epsilon-greedy strategies, ensure robust model performance. Results show a 15% improvement in quiz scores, a 20% increase in knowledge retention, and enhanced engagement metrics, with 80% of system actions yielding positive outcomes. This study highlights the potential of DQN to advance AI-driven education and foster equitable, personalized learning experiences.