Harnessing Generative Artificial Intelligence and Retrieval-Augmented Generation for Advanced Software Engineering, Automated Code Synthesis, and Intelligent Debugging

Abstract

Generative artificial intelligence (AI), particularly large language models (LLMs) trained on code, has quickly become part of
everyday software engineering. These models can write functions from natural-language descriptions, explain unfamiliar code,
and suggest fixes for defects. Used on their own, however, they often produce code that ignores project-specific conventions,
calls nonexistent interfaces, introduces security weaknesses, or misdiagnoses bugs. Retrieval-augmented generation (RAG)
addresses part of this problem by grounding generation in relevant project knowledge such as source files, interface definitions,
documentation, and past defects. This article develops an integrated framework for applying generative AI and RAG to code
synthesis and debugging in enterprise settings. Using a design-oriented approach grounded in ten studies published between
2016 and 2023, the framework links knowledge indexing, context retrieval, generation, automated verification, iterative
debugging, and governed approval workflows. The article discusses how the framework applies to distributed service-based
systems and resource-constrained monitoring applications, and illustrates it with a debugging scenario in a smart monitoring
platform. It argues that generative AI delivers reliable engineering value only when generation is grounded in project context,
checked by automated verification, and governed by structured review.