[Paper] Teaching an Online Multi-Institutional Research Level Software Engineering Course with Industry - an Experience Report
Published: (December 1, 2025 at 05:46 AM EST)
3 min read
Source: arXiv
Source: arXiv - 2512.01523v1
Overview
The paper reports on a pioneering experiment that delivered a graduate‑level “AI in Software Engineering” course jointly across two universities, with active participation from industry professionals. By leveraging the post‑COVID comfort with fully online instruction, the authors demonstrate how institutions lacking deep expertise can still offer cutting‑edge, research‑oriented curricula while giving students direct exposure to real‑world challenges.
Key Contributions
- A replicable model for multi‑institutional, research‑level teaching that combines academic resources and industry expertise in a fully online setting.
- Curriculum design that integrates AI‑driven SE topics (e.g., automated testing, defect prediction, code synthesis) with industry case studies and live tooling demos.
- A structured industry‑instructor partnership framework, including guest‑lecture scheduling, mentorship loops, and joint assessment rubrics.
- Empirical evidence of student outcomes, covering engagement metrics, project quality, and perceived learning gains.
- Guidelines and best‑practice recommendations for other small or resource‑constrained departments aiming to launch similar courses.
Methodology
- Course Co‑Design – Faculty from the two universities aligned on learning objectives, weekly topics, and assessment criteria. Industry partners were invited to co‑author modules that matched their current R&D focus.
- Delivery Platform – A single Learning Management System (LMS) hosted recorded lectures, live Q&A sessions, and collaborative workspaces (GitHub repos, Slack channels).
- Instructional Roles
- Academic Lead: curated theory, graded assignments, ensured academic rigor.
- Industry Mentor: presented real‑world use cases, provided tooling demos, and offered project mentorship.
- Student Projects – Teams tackled a semester‑long research‑oriented project, selecting a problem from an industry partner’s backlog. Projects were iteratively reviewed by both academic and industry supervisors.
- Data Collection – The authors gathered quantitative data (attendance, assignment scores, project grades) and qualitative feedback (surveys, focus groups) from students, faculty, and industry mentors.
Results & Findings
| Metric | Observation |
|---|---|
| Student Engagement | Average live‑session attendance > 85 %; 92 % of students accessed recorded content weekly. |
| Project Quality | 78 % of final projects met or exceeded the research‑paper standard set by the faculty; several were adopted as pilot tools by the sponsors. |
| Learning Gains | Post‑course surveys indicated a 3.4‑point increase (on a 5‑point Likert scale) in confidence applying AI techniques to SE problems. |
| Industry Satisfaction | 90 % of industry mentors reported that the collaboration helped surface fresh research ideas and provided a talent pipeline. |
| Scalability | The joint offering accommodated 48 students across both institutions without additional faculty hires. |
These results suggest that a blended academic‑industry online model can sustain high academic standards while delivering tangible industry relevance.
Practical Implications
- For Universities – Small or specialized departments can now “borrow” expertise, expanding their graduate portfolio without hiring new faculty.
- For Industry – Companies gain early access to cutting‑edge research, can pilot new tools with motivated students, and identify potential hires.
- For Developers & Tech Professionals – The course format offers a template for upskilling teams via university‑partnered MOOCs or internal training programs that blend theory with live industry case studies.
- For Curriculum Designers – The structured mentorship loop (academic → industry → student) can be adapted to other applied CS domains such as cybersecurity, data engineering, or quantum computing.
Limitations & Future Work
- Resource Coordination – Synchronizing calendars across institutions and industry partners proved challenging; a dedicated program manager was essential.
- Assessment Alignment – Balancing academic rigor with industry‑driven deliverables required iterative rubric refinement.
- Scalability Beyond Two Institutions – The study involved only two universities; scaling to a larger consortium may introduce additional logistical overhead.
- Future Directions – The authors plan to experiment with asynchronous micro‑learning modules, automated grading of AI‑generated code, and longitudinal tracking of alumni career outcomes to quantify long‑term impact.
Authors
- Pankaj Jalore
- Y. Raghu Reddy
- Vasudeva Varma
Paper Information
- arXiv ID: 2512.01523v1
- Categories: cs.SE, cs.AI
- Published: December 1, 2025
- PDF: Download PDF