A case study on using customized large language models for requirement-to-code transformation in software engineering
Journal of Emerging Innovations in Engineering · 30.06.2026 · 2 (02) · 88-94
DOI: 10.65664/jeie.v2i02.22 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Emerging Innovations in Engineering, online erschienen am 30.06.2026. Verfasst von Seemal Shahbaz. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Large Language Models (LLMs) have shown great promise for assisting software engineering tasks that can be automated, like mapping natural language to code. This paper describes a practical case study using a domain specific LLM to convert software requirements into semi-formal designs, tests and code. We suggest a simple iterative refinement process for engineering prompts and progressively prompting with the intention of improving the accuracy of outputs, alignment to software requirements. We validate our approach with a case study based on real-world scenarios that show how LLMs can be leveraged for requirement extraction, Generation of Object-oriented Design (OOD) and implementation steps. These findings also give an understanding of the utility of LLM-assisted development in relation to affordances and constraints.
Zitieren
APA
Shahbaz, S. (2026). A case study on using customized large language models for requirement-to-code transformation in software engineering. Journal of Emerging Innovations in Engineering, 2(02), 88-94. https://doi.org/10.65664/jeie.v2i02.22
Vancouver
Shahbaz S. A case study on using customized large language models for requirement-to-code transformation in software engineering. Journal of Emerging Innovations in Engineering. 2026;2(02):88-94. doi:10.65664/jeie.v2i02.22
BibTeX
@article{Shahbaz2026,
author = {Shahbaz, Seemal},
title = {A case study on using customized large language models for requirement-to-code transformation in software engineering},
journal = {Journal of Emerging Innovations in Engineering},
year = {2026},
volume = {2},
number = {02},
pages = {88--94},
doi = {10.65664/jeie.v2i02.22}
}
EUVIMED