Year: 2025
Type: article
Abstract: Building scalable and reliable machine learning models is critical for cloud-native AI systems. Kubeflow provides a robust framework for orchestrating model development workflows. This article presents best practices for ML model development in Kubeflow Cloud-Native Systems, with a focus on Azure Kubernetes Service environments. It explores strategies for optimizing cluster configuration, designing modular and reproducible training pipelines, and implementing effective model tracking and versioning processes. Real-world case studies highlight practical applications of these techniques. The article further evaluates results through system performance metrics and model development outcomes and concludes by discussing lessons learned and future trends in cloud-native ML system design.
Authors: Yevhen Bershchanskyi, Oleksandr Stepanov
Institution: Lviv Polytechnic National University
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W4411231335