ARCHITECTURE DESIGN PATTERNS FOR MACHINE LEARNING SYSTEMS

Abstract

The article provides an extensive systematic review of architectural design patterns for machine learning systems
(ML systems) used in solving predictive analytics problems in dynamic business processes and growing data volumes.
The feasibility of using architectural patterns as formalized practices that ensure structuring, scalability, reproducibility
of experiments, and model lifecycle management is substantiated. A multi-level classification of patterns is proposed by
the stages of the machine learning lifecycle (data collection, cleaning, and preparation, feature engineering, training and
validation, deployment, monitoring, retraining) and by system architectural levels (data layer, model layer, application
and infrastructure layers). The functional purpose, benefits, potential risks, and limitations of the application of key
patterns are analyzed, in particular, Data Pipeline, Feature Store, Training–Serving Split, Model Registry, Microservices,
Event-Driven Architecture, as well as MLOps practices. Their role in ensuring data consistency between training and
production environments, managing model and dataset versions, automating CI/CD processes, organizing forecast quality
monitoring, and detecting model degradation is revealed. Special attention is paid to the issues of integrating ML
components into corporate information systems, ensuring fault tolerance, and supporting a continuous feedback-based
model improvement cycle. Practical recommendations are formulated for selecting and combining patterns, taking into
account the specifics of the subject area, performance requirements, infrastructure scale, regulatory constraints, and the
level of maturity of the development team. The results obtained can be used when designing adaptive and scalable
architectures of predictive analytics systems of various levels of complexity.
Keywords: machine learning, software systems architecture, design patterns, predictive analytics, MLOps.

References
1. Kreuzberger D., Kühl N., Hirschl S. Machine Learning Operations (MLOps): Overview, Definition, and
Architecture // IEEE Access. 2023. Vol. 11. P. 12345–12359. https://doi.org/10.1109/ACCESS.2023.3280048
2. Eken B., Pallewatta S., Tran N. K., Tosun A., Babar M. A. A Multivocal Review of MLOps Practices,
Challenges and Open Issues // arXiv. 2024. arXiv:2406.09737. https://arxiv.org/abs/2406.09737
3. Ганчук Д. О., Семеріков С. О. Інженерія MLOps: метасинтез інструментів, практик та архітектур для
автоматизації машинного навчання // Ukrainian Journal of Information Systems and Data Science. 2024. No. 3(14). P.
45–62. https://jujisds.donnu.edu.ua/article/view/17343
4. Heiland L., Hauser M., Bogner J. Design Patterns for AI-based Systems: A Multivocal Literature Review and
Pattern Repository // Proceedings of the IEEE/ACM International Conference on AI Engineering (CAIN). 2023. P. 184–
196. https://doi.org/10.1109/CAIN58948.2023.00029
5. Amou Najafabadi F., Bogner J., Gerostathopoulos I., Lago P. An Analysis of MLOps Architectures: A
Systematic Mapping Study // European Conference on Software Architecture (ECSA). 2024. P. 69–85.
https://doi.org/10.1007/978-3-031-70797-1_5
6. Zarour M., Alzabut H., Al-Sarayreh K. T. MLOps Best Practices, Challenges and Maturity Models: A
Systematic Literature Review // Information and Software Technology. 2025. Vol. 170. Article 107733.
https://doi.org/10.1016/j.infsof.2024.107733
7. Microsoft Azure Architecture Center. Machine Learning Operations (MLOps): Architectural Patterns and
Pipelines // Microsoft Learn. 2024. [Electronic resource]. https://learn.microsoft.com/azure/architecture/aiml/guide/mlops-technical-paper
8. Sculley D., Holt G., Golovin D. et al. Hidden Technical Debt in Machine Learning Systems // Advances in
Neural Information Processing Systems (NeurIPS). 2015. Vol. 28. P. 2503–2511.
9. Breck E., Cai S., Nielsen E., Salib M., Sculley D. The ML Test Score: A Rubric for ML Production Readiness
and Technical Debt Reduction // Proceedings of the IEEE International Conference on Big Data. 2017. P. 1123–1132.
10. Washizaki H., Uchida H., Khomh F., Gueheneuc Y.-G. Studying Software Engineering Patterns for Machine
Learning Systems // Journal of Systems and Software. 2022. Vol. 188. Article 111239.
https://doi.org/10.1016/j.jss.2022.111239