Industrial organizations are rapidly adopting cloud technologies to reduce infrastructure management costs and increase operational flexibility. Nonetheless, existing cloud deployment uses are mainly focused on infrastructure as a service and data storage. In parallel, the demand for data-driven predictive modeling has grown and continues to evolve beyond standard business intelligence. As a result, cloud-native software development and deployment models have attracted significant attention from academia and industry. However, related research into the cloud-native paradigm remains limited and tends to offer design and development support for a single application or service without addressing the complete data ecosystem.

The data processing and predictive modeling lifecycle requires a cloud-native architecture that supports large-scale predictive modeling operations and the construction of a data-oriented infrastructure. Building intelligent capabilities involves orchestrating all machine learning operations in the cloud, integrating advanced machine learning lifecycle frameworks, incorporating feature stores, and enabling real-time scoring orchestration and execution algorithms. A predictive decision-making data ecosystem focuses on the complete data processing pipeline in relation to predictive models, integrating model data processing and inference pipelines into orchestration and scheduling frameworks while supporting event-driven workflows.