Against the dual backdrop of all-round iteration of artificial intelligence and independent breakthroughs in the Information Technology Application Innovation (ITAI) Industry, the training system for compound practice-oriented talents majoring in computer science under emerging engineering education urgently requires systematic transformation and upgrading. Currently, computer science classrooms in application-oriented undergraduate universities generally face structural problems such as difficulties in implementing precision teaching, insufficient support for personalized adaptive learning, rigid traditional teacher-student dualistic teaching relationships, and fragmented high-level innovative education systems. These manifest prominently as the disconnection between technical training and value guidance, the misalignment between knowledge instruction and innovation ability cultivation, the detachment of teaching models from industrial technology iteration, and the imbalance between single summative assessments and full-process education traceability, making it difficult to meet the urgent talent supply demands for national computing infrastructure construction and the independent controllability of the ITAI industrial chain.

Grounded in the fundamental task of fostering virtue through education in Emerging Engineering, aligned with the development trend of generative AI empowering the digital transformation of education, and relying on the hierarchical advancement theory of AI-empowered classroom teaching, this paper integrates AI large models, supercomputing technology, and smart teaching concepts to originally construct a "Five-in-One, Closed-Loop Iterative" AI-empowered new model for computer science classroom teaching. By developing discipline-specific AI agents, constructing a three-dimensional curriculum map of knowledge-ability-problem, and creating a new ubiquitous education scenario featuring "Teacher-Student-Machine" ternary collaboration, this model reconstructs a full-cycle smart education ecosystem encompassing intelligent pre-class preparation, collaborative in-class teaching, precise post-class enrichment, and intelligent full-process assessment, forming an integrated teaching practice paradigm that is "interdisciplinary, multi-platform, and adaptive," thereby achieving a fundamental transformation in the relationship between classroom teaching and learning. Synchronously, it establishes a four-dimensional education chain of "Curriculum Foundation—Project Practice—Competition Enhancement—Certification Empowerment," systematically overcoming persistent educational reform challenges such as superficial ideological and political infiltration in engineering classrooms, fragmented innovation cultivation, formalistic industry-education integration, and superficial intelligent empowerment. Parallel controlled teaching practices across multiple universities demonstrate that this model can effectively consolidate students' practical abilities in ITAI engineering, deeply cultivate their professional ethics of serving the country through science and technology, and temper their independent breakthrough innovative thinking. Simultaneously, it assists teachers in intelligent workload reduction, precision teaching, and quality and efficiency improvement, offering a replicable, low-cost, and standardized practical paradigm for the large-scale implementation of teaching reform in computer science majors at application-oriented undergraduate universities.