Day 79 of 133

Embeddings + vector DBs + ANN indexes

MTEB; pgvector vs Pinecone vs Weaviate vs Qdrant; HNSW vs IVF.

DSA · NeetCode 1-D DP

Partition Equal Subset SumDSA · 1-D DP
Interview questions to prep
1. State the DP: define the state, the transition, and the base case explicitly.
2. Top-down (memoized recursion) vs bottom-up (tabulation) — which is more natural here, and why?
3. Can you space-optimize from O(n) to O(1)? Show the rolling-window trick.

Embedding models: ada, e5, bge, voyage, jinaGenerative AIMTEB
Interview questions to prep
1. How do you pick an embedding model — what does the MTEB leaderboard tell you and not tell you?
2. Why does dimension count matter for cost and recall?
Vector DBs: Pinecone, Weaviate, Qdrant, pgvectorGenerative AIPinecone
Interview questions to prep
1. Compare Pinecone, Weaviate, Qdrant, and pgvector — when does each fit?
2. When is pgvector enough vs when do you need a dedicated vector DB?
ANN indexes: HNSW, IVF, ScaNNGenerative AIFAISS
Interview questions to prep
1. Compare HNSW vs IVF for ANN — accuracy/speed/memory trade-offs.
2. When would you switch from in-memory FAISS to a hosted vector DB — what's the breakpoint?
Sentence-BERT vs vanilla BERT embeddingsGenerative AISentence-BERT
Interview questions to prep
1. Why does mean-pooled vanilla BERT often underperform Sentence-BERT for semantic search?
2. When is cosine similarity insufficient for retrieval quality, and what diagnostics would you add?
3. How do you evaluate embedding quality beyond a single nearest-neighbor demo?

References & further reading