Home / GenAI Engineer practice test / Design Applications / Question

Databricks Generative AI Engineer Associate · free practice question

A product manager at a retail firm outlines a new AI capability: 'On a customer's order history screen, display three recommended items we believe they'll…

Exam-style question on Design Applications from the free GenAI Engineer practice test, with the answer and a full explanation below.

Updated November 11, 2026

Design Applications

A product manager at a retail firm outlines a new AI capability: 'On a customer's order history screen, display three recommended items we believe they'll appreciate, accompanied by a brief justification explaining the personal relevance of each item.' An AI pipeline engineer needs to convert this high-level business concept into a precise input/output specification. Which of the following specifications MOST accurately reflects this requirement?

  1. AInput: a static CSV of the product catalog sorted by overall popularity. Output: the three best-selling products as a numbered list of names.
  2. BInput: the customer's purchase history plus the product catalog or a vector index of it. Output: three products, each with product_name and a short reasoning.
  3. CInput: the product reviews the customer has written since opening the account. Output: one summary paragraph describing the customer's overall preferences.
  4. DInput: the customer's demographic attributes such as age bracket and postal code. Output: three products popular with that demographic group, without a rationale.
  5. EInput: the embedding of the customer's most recent purchase. Output: cosine similarity scores for the 50,000 catalog items, sorted so the top three are the picks.
Answer & explanation

Correct answer: B

WHY B is correct: A good specification maps the business goal to concrete inputs and structured outputs. Personalization needs the customer's purchase history together with the catalog to choose from, and the request for a brief justification per item means the output must carry a natural-language reasoning for each of exactly three products, not just identifiers.

WHY NOT A: A global bestseller list ignores the personalization requirement: every customer would see the same three products whatever their history. Returning names alone also leaves out the explanation of why each product suits this customer.

WHY NOT C: A paragraph about the customer's tastes is not a recommendation. The product manager asked for three specific products; a taste profile may be a useful intermediate step in the pipeline, but it is not the final output described.

WHY NOT D: Demographic targeting is not personalization from the customer's own behavior: it does not use what this customer actually bought. The output also omits the personal explanation the requirement calls for.

WHY NOT E: Raw cosine similarity scores are an intermediate computation, not something to show on a customer's screen. The requirement is a named product with a readable explanation, and using a single recent purchase discards the wider purchase history needed for good personalization.

Take the full GenAI Engineer practice test free →

Would you pass today?

This question is one of 45 in the free, timed Databricks Generative AI Engineer Associate practice test. Take it and get a section-by-section score.

Start the free test →