MLOps

The article investigates the problem of increasing the efficiency of the data labeling process in poorly controlled environments based on Active Learning methods. The relevance of the work is due to the rapid growth of unstructured and partially labeled data, the high cost of manual annotation, the shortage of qualified experts, and the negative impact of noise labels on the quality of machine learning models.

This article presents the main findings from an in-depth study of MLFlow design concepts in containerized and cloud-native systems. The research focuses on how MLFlow, as a core MLOps framework, can be efficiently deployed and managed in cloud- native environments to ensure scalable, reproducible, and secure machine learning workflows. The study analyzes the architectural principles and integration patterns of MLFlow components: Tracking Server, Projects, Models, and Registry within distributed containerized infrastructures.