Databricks-GenAI-Assoc Design Applications Practice Question
A generative AI engineer is designing a Databricks RAG application that ingests a Delta table containing 400 million support articles into a Databricks Vector Search index. The team wants the lowest-latency online serving with the smallest possible index while preserving retrieval quality for the most common queries. Which design decision best meets these requirements?
⚠ Common exam trap
The trap here is assuming that a RAG application must always embed every row of the source table, when pre-filtering the Delta table before the Delta Sync index is created is a valid way to shrink the index and improve latency.
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
✓
Use Databricks Vector Search with a Delta Sync index, and configure a smaller index by filtering out low-traffic or stale articles before syncing so only high-value documents are embedded.
Pruning low-value content before building a Delta Sync index directly reduces the number of embeddings stored and searched, which shrinks the index and lowers serving latency. Because Vector Search Delta Sync indexes continue to track the source Delta table, the team retains freshness and can still serve high-quality results for the queries that matter most, matching both the performance and quality goals.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Store all 400 million embeddings in a Unity Catalog volume and query them directly with a Python UDF at request time.
Why it's wrong here
Querying raw embedding files in a Unity Catalog volume with a UDF bypasses the managed Vector Search serving layer, so there is no optimized approximate nearest neighbor index and no low-latency endpoint. Scanning hundreds of millions of vectors per request would be far slower and more expensive than the managed index, and it does not satisfy the smallest-index requirement.
- ✗
Replace Databricks Vector Search with a keyword-only BM25 index over the raw article text to avoid embedding storage costs.
Why it's wrong here
Dropping vector search for keyword-only retrieval abandons semantic matching, which is central to RAG quality on natural-language support questions. The scenario asks for retrieval quality preservation for common queries; a BM25-only design will miss paraphrased or conceptual matches. It also does not produce a vector index, so the stated latency and design goals for the RAG application are not met.
- ✓
Use Databricks Vector Search with a Delta Sync index, and configure a smaller index by filtering out low-traffic or stale articles before syncing so only high-value documents are embedded.
Why this is correct
Filtering the source Delta table to high-value, frequently accessed articles before the Delta Sync index is built reduces the number of embeddings stored and searched, which lowers latency and index size. Vector Search Delta Sync indexes stay current with the source table, so the team can prune low-traffic content while keeping retrieval quality for the common queries the business cares about.
- ✗
Enable the vector index with the vector search endpoint sized for the full 400 million rows and keep all embeddings in a single index.
Why it's wrong here
Sizing a single endpoint to hold all 400 million rows maximizes memory and cost without addressing the stated goal of a smaller index. The scenario asks for the smallest possible index while preserving quality for common queries; keeping every row in one index ignores that opportunity to prune or tier content and increases serving latency and expense unnecessarily.
Visual reference
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.