A Unified Machine Learning Approach for Efficient ArtifactManagement in Jenkins CI/CD Pipelines

Abstract

In this paper, we provide an innovative way to manage artifacts in Jenkins-based CI/CD pipelines using a single flexible ML model. Managing artifacts effectively is a priority when teams grow. In this paper, we present a single ML model that can handle various artifact management problems such as retention prediction, compression optimization, artifact classification, cache optimization, and anomaly detection. With just one model deployed across these multiple functions, we are able to simplify things, reduce compute overhead, and provide a highly scalable solution for Jenkins deployments on large scale Among the major findings of this study are:Creation of a generic multi-task learning system that helps to manage artifacts in CI/CD pipelines. Implementation of the model to solve multiple artifact management problems in a holistic manner.Easy integration of the ML-based solution into Jenkins processes, enabling automation and efficiencies.Performance testing in enterprise CI/CD systems to prove scalability and effectiveness. Comprehensive review of benefits and limitations of artifact management using a single ML model.This work pushes the field forward with a single and complete solution, providing scalability, resource-savings, and artifact lifecycle management in large scale, CI/CD ecosystems.