A compliance team requires every deployed AI model to have documented lineage showing which data sources were used for training and which code produced the model. The team uses Unity Catalog. Which combination of Databricks features addresses this requirement most completely?

Question

Accepted Answer

B. Registering models to Unity Catalog (which captures model lineage) combined with MLflow experiment tracking (which logs training run details including data and code references).. Unity Catalog captures model lineage as part of model registration, associating model versions with the MLflow experiment runs that produced them and the data assets referenced. MLflow experiment tracking captures the code version, parameters, and dataset references for each training run. Together they provide end-to-end automated lineage from data to deployed model.

A compliance team requires every deployed AI model to have documented lineage showing which data sources were used for training and which code produced the model. The team uses Unity Catalog. Which combination of Databricks features addresses this requirement most completely?

Related Questions

Discussion