Intelligent CI Failure Analysis with AI Systems

Abstract

This thesis presents the research, design and development of a tool that provides intelligent analysis of Continuous Integration (CI) systems in the Software Development Lifecycle (SDLC) that aims to provide the product development teams the required assistance to troubleshoot issues related to CI logs utilizing Large Language Models (LLMs) providing failure analysis with Root Cause Analysis (RCA) to mitigate product line issues early from problem to solution. Thus, the proposed tool is intended to reduce the effective time spent by the developer to analyse the failure.

The foundations of the system integrate with the SDLC to support early detection of product development issues by using Artificial Intelligence (AI) to analyse CI pipeline logs, version control information and contextual data from locally hosted LLMs. The study explores an efficient troubleshooting environment that strengthens reliability in the CI workflow and potentially increases the engineers effectiveness