Your RAG keeps retrieving near-duplicate chunks, wasting the context window. How do you diversify results?
Top-k by raw similarity often feeds the model five paraphrases of one paragraph, denying it the other facts the question needs. The fix is ranking for relevance and novelty together. Here is how.
Updated Sep 2026 · Grounded in real GenAI, LLM, and AI/ML engineering interview loops and written to a senior-engineer editorial bar.
Top-k by raw similarity often feeds the model five paraphrases of one paragraph, denying it the other facts the question needs. The fix is ranking for relevance and novelty together. Here is how.
Lead with where the obvious approach breaks, because that is the judgment they are screening for — most candidates jump straight to the happy path and lose the room.
Then walk the failure back through the pipeline in order, naming the one metric the customer's exec sponsor actually cares about before you propose the fix.
