PersonalPlan: Planning Multi-Agent Systems for Personalized Programming Learning

2026-06-16

AI Summary

This study addresses the limitations of existing large language model (LLM)-driven multi-agent systems in programming education, which often lack awareness of learner profiles and pedagogical scaffolding, thereby failing to deliver personalized support. To overcome this, the authors propose a multi-agent planning framework that integrates learner profiling with instructional dependency modeling to generate executable, personalized, and pedagogically coherent learning plans through a two-stage training process. The approach employs hierarchical supervised fine-tuning combined with LoRA adapters to capture task decomposition and step-wise dependencies, and introduces a reward-adaptive GRPO algorithm to optimize plan quality. Evaluated on the newly curated MAP-PPL dataset comprising 3,043 instances, the method significantly outperforms current LLM and multi-agent baselines at both 8B and 32B scales, achieving state-of-the-art performance in plan executability, personalization, and instructional quality.

Abstract

Effective programming education requires personalized instruction adapted to diverse learner backgrounds. However, while LLM-based multi-agent systems (MAS) excel at complex planning, existing planners often lack profile-grounding and pedagogical scaffolding, thereby undermining personalized programming learning. To fill in the gap, we first introduce \textbf{MAP-PPL} (\textbf{M}ulti-\textbf{A}gent \textbf{P}lans for \textbf{P}ersonalized \textbf{P}rogramming \textbf{L}earning), a profile-conditioned multi-agent planning dataset with 3{,}043 query--profile--plan instances from 1{,}730 Stack Overflow question groups and 2{,}738 learner profiles. Each plan specifies agents, subtasks, executable steps, and prerequisite dependencies. Then, we propose \textbf{PersonalPlan}, a two-stage MAS planner that first performs hierarchical SFT with separate LoRA adapters for profile-aware task decomposition and step dependency planning, then applies a Reward-Adaptive GRPO to encourage the model to generate executable, personalized, and pedagogically scaffolded plans. Extensive experiments on MAP-PPL comparing PersonalPlan against frontier LLMs, generic MAS frameworks, and agentic planners demonstrate its superiority. With only 8B and 32B variants, PersonalPlan achieves state-of-the-art plan executability, personalization, and pedagogical quality, effectively orchestrating MAS for agent-student interactions.

Problem

• personalized programming learning
• multi-agent systems
• profile-grounding
• pedagogical scaffolding
• LLM-based planning

Innovation

• Personalized Programming Learning
• Multi-Agent Planning
• Profile-Grounded Planning
• LoRA-based Hierarchical SFT
• Reward-Adaptive GRPO