Question

A Generative AI Engineer is packaging a RAG chain for deployment to a Databricks Model Serving endpoint. The chain queries a Databricks Vector Search index and calls a Foundation Model API endpoint at inference time. During local testing the chain works, but after deployment the served endpoint fails with authentication errors when trying to reach the Vector Search index. What is the correct way to ensure the served model can securely access these Databricks resources without embedding personal credentials?

Accepted Answer

Specify the Vector Search index and the serving endpoint in the `resources` parameter when logging the model with `mlflow.langchain.log_model()`, so Databricks provisions automatic authentication (on-behalf-of credentials) for the endpoint.

Answer

Hardcode a personal access token (PAT) into the chain code before logging the model so the deployed endpoint can authenticate to Vector Search.

Answer

Add the Databricks workspace URL and a PAT as environment variables at log time using `mlflow.set_tag()` so they are baked into the model artifact.

Answer

Grant the endpoint public network access to the Vector Search index and disable authentication on the index to avoid credential issues.