Prompt Engineering in Software Engineering Education: An Empirical Study of Demand, Supply, and Assessment

Abstract

Prompt engineering is increasingly embedded in software development practice, yet software engineering curricula lack evidence-grounded guidance on which prompt-related competencies to prioritize and how to assess them. This paper presents a large-scale empirical triangulation of industry demand and curricular supply for prompt-engineering competencies in software development. By analysing prompt-related job advertisements across major economies and auditing course descriptions from leading computer science programs, we identify systematic mismatches between practice-facing expectations and formal educational coverage. Industry demand consistently emphasizes prompt design alongside evaluation and testing, with strong signals for prompt optimization and a substantial subset of postings highlighting retrieval augmentation and orchestration. In contrast, university curricula predominantly foreground prompt creation and refinement, with limited explicit emphasis on systematic evaluation. Building on this triangulated evidence, we distil three recurring gaps and derive assessment-oriented implications, specifying inspectable student artefacts, such as evaluation harnesses, lifecycle documentation, and safety testing evidence, that allow prompt-engineering competence to be evaluated in software engineering terms. Collectively, the study provides an empirical foundation to inform curriculum design and assessment in software engineering education amid the rapid diffusion of large language models.