GenAI Engineer mock exam

Question 1, Design ApplicationsBusiness users ask questions about internal PDFs, and answers must cite sources. What should you design?

• AA nightly BI dashboard built from the extracted PDF text
• BA plain classification model that labels each question
• CA RAG application with citations in the output
• DA fine-tuned model with no retrieval step

Question 2, Data PreparationAnswers lose context because long sections were cut at a fixed character count. What do you change?

• AThe LLM's temperature, so it fills gaps in the context
• BThe chunking strategy, so it respects document structure
• CThe Unity Catalog owner of the index
• DThe serving endpoint size and its scale-out limit

Question 3, Data PreparationDocuments change daily, and the vector index must stay current with little custom work. What fits?

• AA scheduled notebook that rebuilds the whole index
• BAn AI Search index built from a Delta table
• CEmbeddings stored in a CSV and reloaded nightly
• DA form that asks users to reupload changed files

Question 4, Application DevelopmentAn assistant must look up live order status during a conversation. What do you add?

• AA tool the agent can call, backed by the order system
• BMore documents added to the vector index
• CA larger embedding model for the order data
• DA longer chunk size on the order history

Question 5, Application DevelopmentYou need the app to return consistent JSON for a downstream system. What is the best approach?

• ARaise the temperature and retry on failure
• BParse the free text with a long regular expression
• CAsk for JSON politely in the prompt and trust what comes back
• DUse structured output with a defined schema and validate it

Question 6, Assembling and Deploying AppsYou want to package a chain, log its versions, and deploy it behind an endpoint. Which pair fits?

• ADelta Sharing and a dashboard
• BA notebook and a shared drive
• CGit and email
• DMLflow and Model Serving

Question 7, GovernanceDifferent teams should only reach the models and tables they own. Where do you enforce this?

• AIn a prompt instruction for each agent
• BIn a naming convention for each team
• CIn Unity Catalog permissions
• DIn the title of each notebook

Question 8, Evaluation and MonitoringAfter launch, you want to catch quality problems from real traffic. What do you set up?

• AA second vector index kept as a warm standby
• BA once-a-year review of the prompts and the model
• CMonitoring and logging of requests, with evaluation on samples
• DA bigger cluster to handle the extra traffic