A Generative AI Engineer is tasked with deploying an application that takes advantage of a custom MLflow Pyfunc model to return some interim results. How should they configure the endpoint to pass…

Question

Options

• AUse spark.conf.set ()
• BPass variables using the Databricks Feature Store API
• CAdd credentials using environment variables
• DPass the secrets in plain text

• A
  13% (6)
• B
  7% (3)
• C
  76% (35)
• D
  4% (2)

Explanation

When deploying an MLflow Pyfunc model as a Databricks Model Serving endpoint, the correct and secure way to pass secrets and credentials is through environment variables. Databricks Model Serving allows you to configure environment variables (backed by Databricks Secrets) for your endpoint so that the model code can read them at runtime without exposing sensitive values. Option A (spark.conf.set()) is used to configure Spark session properties, not model serving endpoints. Option B (Databricks Feature Store API) is for retrieving feature data for inference, not for credential management. Option D (plain text) is never acceptable for secrets because it exposes credentials in logs, code, and configuration files, violating basic security principles.

No community discussion yet for this question.