This study proposes an AI-supported project-based learning (PBL) framework for the Data Structures course under the Less Teaching, More Learning philosophy. The traditional chapter-based curriculum was reconstructed into eight authentic Python projects supported by the learning activity management system and a large language model-based AI teaching assistant. A controlled teaching experiment involving two undergraduate classes was conducted to evaluate the effectiveness of the proposed approach. The results indicate that the AI-supported PBL model significantly improved students’ academic performance, programming competence, and self-directed learning compared with conventional instruction. The proposed framework provides a practical and transferable paradigm for AI-enabled curriculum reform in computer science education.