Instruction: Explain how you would debug a RAG system that struggles with document structure.
Context: Tests how the candidate diagnoses the problem, chooses the safest next step, and reasons through recovery. Explain how you would debug a RAG system that struggles with document structure.
Official answer available
Preview the opening of the answer, then unlock the full walkthrough.
I would start before the model. Tables and footnotes are often lost in parsing, flattening, or chunking long before retrieval begins. So I would inspect the document extraction pipeline first: OCR quality, table preservation, section ordering, page breaks, and whether footnotes stay attached to the claims they qualify.
Then I would look...
easy
easy
easy
easy
easy
easy