Databricks-GenAI-Assoc Evaluation and Monitoring Practice Question
A Databricks Generative AI engineer has deployed a RAG application and is now setting up production monitoring. They want to automatically detect when the distribution of incoming user questions diverges from the distribution seen during development, so they can trigger retraining or prompt adjustments. Which Databricks capability should they configure?
⚠ Common exam trap
The trap here is assuming that any monitoring feature can detect input drift, when only inference tables combined with Lakehouse Monitoring profile the actual request payload distribution.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Inference tables with Lakehouse Monitoring for the endpoint's payload and response columns
Inference tables capture the actual request and response payloads served by the endpoint, and Lakehouse Monitoring can build profiles and drift metrics over those tables. Together they provide the automated distributional comparison needed to detect when live questions diverge from the development baseline, which is exactly the monitoring goal.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Mosaic AI Agent Evaluation with a custom LLM judge for every request
Why it's wrong here
Agent Evaluation assesses response quality against a labeled or synthetic evaluation set, typically in batch or scheduled runs. It is not designed to continuously profile the distribution of live user inputs for drift. Using it on every request would also be prohibitively expensive and slow, so it fails the automation requirement.
- ✗
Delta Live Tables expectations on the raw question stream
Why it's wrong here
DLT expectations enforce data quality rules such as non-null or range constraints on a streaming table. They can flag malformed records but do not compute distributional drift between current and baseline data. This makes them unsuitable for detecting semantic or statistical divergence in user question distributions.
- ✓
Inference tables with Lakehouse Monitoring for the endpoint's payload and response columns
Why this is correct
Inference tables log the request payload and model response for a Mosaic AI Model Serving endpoint, and Lakehouse Monitoring can profile those tables to compute drift metrics on the input distribution. This directly addresses detecting when live user questions diverge from the development baseline, enabling automated alerts and retraining triggers.
- ✗
MLflow Model Registry stage transitions with webhook notifications
Why it's wrong here
Model Registry stages and webhooks track model version promotion events, not the statistical distribution of incoming request data. They cannot detect input drift because they have no visibility into live inference payloads. This capability is about deployment lifecycle, not runtime data monitoring, so it does not satisfy the requirement.
About these practice questions
One of 330 original Databricks-GenAI-Assoc practice questions on Courseiva, each with a full explanation and wrong-answer analysis — not exam dumps or protected exam content. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Databricks exam blueprint
This Databricks-GenAI-Assoc practice question is part of Courseiva's free Databricks certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the Databricks-GenAI-Assoc exam.