Understanding reference architectures for MLOPs

Machine learning jobs are now an essential part of any software and organization. Moreover, machine learning and data analytics components are both dynamic and continuous. Therefore, Machine Learning Operations (MLOps) is crucial for organizations. Figure 10.6 shows the reference architecture for MLOps. This section will discuss how IBM Cloud Pak for Data can provide an end-to-end MLOps solution.

Project governance

MLOps mainly needs project governance, security operations such as identity and access control management, and data and network security operations. MLOps starts with project governance, where organizations can manage multiple data analytics and insight projects. Each project has its assets, such as data, Jupyter notebooks, models, people, dashboards, data flows, and more:

Figure 10.6 – Reference architecture for MLOps

Figure 10.7 shows different assets for a data science project in IBM...