Agentic AI for Personalized Learning

Adaptive tutoring systems aim to personalize instruction by modeling a learner's current mastery and dynamically selecting the next concept to teach.This paper presents an Agentic AI framework for personalized learning that combines transformer-based knowledge tracing, autonomous decision-making, and dynamic content generation.The proposed system estimates student mastery using an attention-based knowledge tracing model, selects the next skill with a Deep Q-Network (DQN), and generates lesson and quiz content for targeted review.The architecture is modular, practical, and suitable for deployment in a web-based tutoring environment.The experimental use of student interaction data and reinforcement-based skill selection demonstrate how adaptive teaching can improve both personalization and learner engagement.