Optimizing the Machine Learning Lifecycle and MLOps: The Emergence of Cloud Data Platforms
To put machine learning models to work, data science teams must manage an ML lifecycle that spans data and feature engineering, model development, and model production. Traditionally, data scientists use ML platforms that incorporate all the capabilities required to prepare data; create features; and train, deploy, and manage models.
Now a new option has emerged: cloud data platforms that merge data warehouse and data lake constructs. Like an ML platform, the cloud data platform offers lifecycle speed, scale of production, model governance, and support for the ecosystem of ML tools. But it also goes further and offers the ability to integrate workflows, collaborate cross-functionally, and consolidate data across both BI and data science projects.