nerdexam
Databricks

GENERATIVE-AI-ENGINEER-ASSOCIATE · Question #93

A Generative AI Engineer is assessing the responses from a customer-facing GenAI application that they are developing to assist in selling automotive parts. The application requires the customer to…

The correct answer is B. Create a feature store table with transaction_id as primary key that is populated with invoice data. Shipping and arrival date questions require real-time, transaction-specific structured data that a feature store table keyed on transaction_id can provide at inference time via lookup.

Building and Optimizing RAG Applications

Question

A Generative AI Engineer is assessing the responses from a customer-facing GenAI application that they are developing to assist in selling automotive parts. The application requires the customer to explicitly input account_id and transaction_id to answer questions. After initial launch, the customer feedback was that the application did well on answering order and billing details, but failed to accurately answer shipping and expected arrival date questions. Which of the following receivers would improve the application's ability to answer these questions?

Options

  • ACreate a vector store that includes the company shipping policies and payment terms for all
  • BCreate a feature store table with transaction_id as primary key that is populated with invoice data
  • CProvide examples data for expected arrival dates as a tuning dataset, then periodically fine-tune
  • DAmend the chat prompt to input when the ordered was placed and instruct the model to add 14

How the community answered

(21 responses)
  • B
    86% (18)
  • C
    5% (1)
  • D
    10% (2)

Why each option

Shipping and arrival date questions require real-time, transaction-specific structured data that a feature store table keyed on transaction_id can provide at inference time via lookup.

ACreate a vector store that includes the company shipping policies and payment terms for all

A vector store of general shipping policies contains static policy text, not the transaction-specific shipping status and expected arrival dates needed to answer questions about a particular order.

BCreate a feature store table with transaction_id as primary key that is populated with invoice dataCorrect

A Databricks Feature Store table with transaction_id as the primary key can be populated with shipping carrier data, estimated delivery dates, and tracking status, then looked up at inference time using the customer-provided transaction_id - delivering the precise, transaction-specific data the model needs to answer shipping questions accurately.

CProvide examples data for expected arrival dates as a tuning dataset, then periodically fine-tune

Fine-tuning on example arrival date data bakes patterns into model weights but does not give the model access to live, transaction-specific shipping records for new orders.

DAmend the chat prompt to input when the ordered was placed and instruct the model to add 14

Hard-coding a 14-day addition to the order date is an inaccurate heuristic that ignores carrier, shipping method, and real-time delays - it will produce wrong answers for most transactions.

Concept tested: Feature store lookup for structured data retrieval in RAG

Source: https://docs.databricks.com/en/machine-learning/feature-store/index.html

Topics

#Retrieval Augmented Generation (RAG)#Feature Stores#Data Retrieval#GenAI Application Design

Community Discussion

No community discussion yet for this question.

Sign up to join the discussion
Full GENERATIVE-AI-ENGINEER-ASSOCIATE Practice