一句话：报告三年UTM三方协作经验，让学生与产业伙伴共同交付生产级AI集成软件，并总结实践挑战与教学策略。
软件工程专业能力要求学生将课堂学习迁移到具有真实约束与利益相关者的职业环境，但传统小组项目、实习和毕业设计往往遮蔽真实客户压力与对生产级软件的共同责任。本文报告三年UTM经验：学生、教师与产业伙伴三方协作，共同交付真实工业级软件，学生参与需求协商、架构决策和部署，而非仅观察职业实践。三年间覆盖50名学生、16个项目及多元合作方，学生既学习使用AI工具提升生产力，也面向AI系统开发智能应用。基于五个案例，作者反思UTM与其他产业参与模式的差异，讨论教育目标与交付压力平衡、技术不确定性下支持学生、跨利益相关者沟通等挑战，并为AI时代设计产业嵌入式协作项目提供可操作策略。
Professional competence in software engineering requires graduates to transfer classroom learning into professional contexts with real constraints and stakeholders. Traditional approaches - group projects, placements, capstones - provide valuable learning but typically shield students from authentic stakeholder pressure and shared accountability for developing production-ready software. This paper reports on three years of experience with the University Talent Marketplace (UTM), a tripartite collaboration model where students, academics, and industry partners jointly deliver real, industry-grade software systems, sharing responsibility for design decisions, implementation, and production-quality outcomes. In UTM, students participate as contributors to the negotiation of requirements, architectural decisions, and deployment, not observers of professional practice. Over this period, UTM engaged 50 students across 16 projects with diverse partners, ranging from multinational corporations to global health initiatives. Students learned to work both with AI tools (for productivity) and for AI systems (building intelligent applications), navigating the dual challenges of AI-assisted development and AI system deployment. Drawing on five case studies, this experience report reflects on what distinguishes the UTM from alternative industry engagement approaches and examines key challenges, including balancing education objectives with delivery pressures, supporting students amid technological uncertainty, and sustaining effective communication across stakeholders. We share practical strategies and insights that offer actionable guidance for educators designing collaborative, industry-embedded programs that develop professional software engineering competence through authentic practice in the AI era.
@inproceedings{NeykovaSISDS26,
title = {Building Production Software and Professional Competence: A Collaborative Model for AI-Integrated Software Engineering Practice},
author = {Rumyana Neykova and Cigdem Sengul and Zear Ibrahim and Isabel Sassoon and Giuseppe Destefanis and Alan Serrano},
booktitle = {FSE},
year = {2026},
doi = {10.1145/3803437.3805795}
}