MLOps I: Feature Store and Model Registry

Interested in operationalizing machine learning models directly where your data lives? Join our Snowflake Solution Engineering team for a hands-on session showing how to go from raw data to governed, production-ready models using Snowflake ML.

In this virtual lab, we’ll walk through an end-to-end workflow using Snowpark, Feature Store, Experiment Tracking, and Model Registry - all inside the Snowflake Data Cloud.

Using fraud detection as a running example, you’ll learn how to:

Build and Compare Models
 

• Engineer features with Snowpark and train a baseline XGBoost fraud model.
• Use Experiment Tracking to log runs and metrics, and quickly compare how different models perform.

Operationalize Behavioral Features with Feature Store
 

• Define Customer and Terminal entities and create various features (recent activity, spend, fraud history) in Feature Store.
• Generate an enriched training dataset, train an improved fraud model, and log multiple versions to Model Registry, setting the best-performing one as the default for downstream use.

By the end of this session, you’ll see how Snowflake ML streamlines feature engineering, model comparison, and deployment - all within a single platform.

Register now to reserve your spot!