RT Journal Article
T1 Towards an open-source MLOps architecture
A1 Burgueño Romero, Antonio Manuel
A1 Benítez-Hidalgo, Antonio
A1 Barba-González, Cristóbal
A1 Aldana-Montes, José Francisco
K1 Soporte lógico libre
AB The need for robust MLOps frameworks has become paramount in a world increasingly reliant on AI. Current solutions range from proprietary cloud services to independent open-source components, each with advantages and limitations. This paper presents a Kubernetes-based, open-source MLOps framework designed to streamline the lifecycle management of machine learning models in production environments. It integrates a comprehensive suite of open-source tools compatible with Python, covering all aspects from development, testing, deployment, and monitoring to updating models, reducing the need for human intervention. Finally, we compared state-of-the-art MLOps tools and frameworks, demonstrating that our framework meets the same features as proprietary options, such as Amazon SageMaker.
PB IEEE Xplore
YR 2024
FD 2024-07-08
LK https://hdl.handle.net/10630/36440
UL https://hdl.handle.net/10630/36440
LA eng
NO A. M. Burgueño-Romero, A. Benítez-Hidalgo, C. Barba-González and J. F. Aldana-Montes, "Toward an Open Source MLOps Architecture," in IEEE Software, vol. 42, no. 1, pp. 59-64, Jan.-Feb. 2025, doi: 10.1109/MS.2024.3421675
DS RIUMA. Repositorio Institucional de la Universidad de Málaga
RD 5 mar 2026