Back to guides
GENAI-ENG Sample Preview
Databricks Generative AI Engineer
200 original LLM application engineering scenarios with answers, explanations, exam tips, and a seven-day revision plan. Preview a few representative questions below before buying the complete protected PDF.
Sample 1. A chatbot must answer from approved company documents. Which pattern is most appropriate?
- Retrieval augmented generation
- Ungrounded free-form generation
- Publicly exposing all documents
- Disabling evaluation
Answer: A - RAG retrieves trusted content before generation, improving groundedness.
Sample 2. A model gives confident but unsupported answers. What is this commonly called?
- Hallucination
- Vector indexing
- Tokenization only
- Schema validation
Answer: A - Hallucination is an unsupported or incorrect generated claim.
Sample 3. What helps compare model quality before release?
- Evaluation set with quality metrics
- Skipping tests
- Only changing logo colors
- Removing user feedback
Answer: A - Evaluation examples and metrics detect regressions and quality issues.
Sample 4. Which control reduces harmful generated output?
- Safety filters and guardrails
- Longer prompts only
- No monitoring
- Public secrets in prompts
Answer: A - Filters and guardrails constrain unsafe or policy-violating outputs.
Sample 5. What converts text into numeric representations for semantic search?
- Embeddings
- Firewall rules
- DNS records
- Invoice exports
Answer: A - Embeddings represent meaning as vectors for similarity search.
