PACER (Personalized Adaptive Coaching for Engagement and Reflection) is an agentic AI tutoring system for IUB courses in Computer Programming |, Computer Programming II and Data Structure. While LLMs can explain, generate, and debug code, standalone use often provides answers without adequately modeling what a learner knows, diagnosing misconceptions, or adapting support over time. PACER builds a learner model from conversations, submitted code, execution behavior, and runtime errors, then uses pedagogical methods to decide what support to provide and when. Rather than simply giving solutions, it guides students toward active problem solving, reflection, and increasingly independent performance while adapting scaffolding as understanding develops. The system will also provide instructors with insights into student progress and difficulties. Using IUB courses as the deployment setting, we will study whether agentic AI can improve learning and engagement while progressively reducing dependence on AI assistance.
Team: Tanvir Mahmud, Amin Ahsan Ali
