Jenkins Pipelines: A Novel Approach to Machine Learning Operations (MLOps)
Dr Manoj Kumar Niranjan, Mohana · 2022 International Conference on Edge Computing and Applications (ICECAA) · 2022
Machine Learning is a widely popular field that is being used in an increasingly large number of projects worldwide. This necessitates the use of certain practices to create a structured framework for such projects. These practices, processes and pipelines are termed as Machine Learning Operations (MLOps). The software development lifecycle for a machine learning is indeed very complex and sequential nature of it results in several repetitive tasks for developers. Automation in this aspect would greatly reduce time and manual effort required. Jenkins is an open-source continuous integration tool that can be used to build pipelines to define and automate workflows in MLOps domain. This paper proposes the design and implementation of pipelines for various stages of MLOps that are data analysis, data preparation, training, testing and deployment on a single platform.