Day 80 of 133

Retrieval strategies: dense / sparse / hybrid / HyDE

BM25 + dense fusion; HyDE; multi-query; multi-hop.

DSA · NeetCode 2-D DP

Burst BalloonsDSA · 2-D DP
Interview questions to prep
1. State the 2-D DP: indices, recurrence, base case. What's the order of fill?
2. Can you reduce 2-D to 1-D by reusing rows or columns? Walk through the dependency direction.
3. Top-down with memoization vs bottom-up — which is easier to reason about, and which is faster in practice?

Dense vs sparse vs hybrid (BM25 + dense)Generative AIPinecone
Interview questions to prep
1. When does sparse retrieval (BM25) beat dense retrieval?
2. How would you combine BM25 and dense scores in hybrid search?
3. Given 30 failed RAG queries, how would you check whether the gold document appears in top-k before touching the generator?
4. When would you add query rewriting, expansion, or hybrid search instead of changing the embedding model?
Query expansion: HyDE, multi-query, query rewritingGenerative AIHyDE
Interview questions to prep
1. How does HyDE work, and when does it actually help?
2. When would you decompose a query into multiple sub-queries?
Dense Passage Retrieval (DPR) and retriever trainingGenerative AIDPR
Interview questions to prep
1. How does DPR train a dense retriever, and how is it different from BM25?
2. What positives and hard negatives would you use to train or fine-tune a retriever?
3. When would DPR-style retrieval fail compared with sparse or hybrid search?
Multi-hop & iterative retrievalGenerative AIFLARE
Interview questions to prep
1. How would you handle a multi-hop question that requires chaining retrieval calls?
2. What stopping criterion do you use for iterative retrieval — and how do you avoid runaway loops?

References & further reading