Skip to content

RAG & Vector Databases interview questions

Embeddings, vector indexing (HNSW, IVFFlat), hybrid search, chunking strategies, re-ranking, and query transformation.

What to study

  • Dense vs sparse embeddings and hybrid keyword + semantic search.
  • Chunking strategies (semantic, sliding window) and document parsing.
  • HNSW vs IVFFlat index tradeoffs for latency, memory, and recall.

Questions and answers

  1. What is the difference between HNSW and IVFFlat vector indexes?

    Intermediate · 2 min read

  2. What is Retrieval-Augmented Generation (RAG) and why is it used?

    Beginner · 2 min read

  3. How do dense and sparse vector embeddings differ?

    Beginner · 2 min read

  4. How does hybrid search combine BM25 and vector embeddings?

    Intermediate · 2 min read

  5. How do cross-encoder rerankers and late-interaction (ColBERT) models work?

    Advanced · 2 min read

  6. What is chunk size and chunk overlap in text splitting?

    Beginner · 2 min read

  7. What is Cosine Similarity vs Euclidean Distance in vector search?

    Beginner · 2 min read

  8. What is the 'Lost in the Middle' phenomenon in LLM retrieval?

    Beginner · 2 min read

  9. How does query rewriting and Hypothetical Document Embeddings (HyDE) work?

    Intermediate · 2 min read

  10. How do GraphRAG and Knowledge Graphs enhance vector retrieval?

    Advanced · 2 min read

  11. What is an embedding model and how does vector dimension affect search quality?

    Beginner · 2 min read

  12. What is semantic search and how does it differ from traditional keyword search?

    Beginner · 2 min read

  13. What is document metadata filtering in vector databases?

    Beginner · 2 min read

  14. What is single-representation vs multi-representation document retrieval?

    Beginner · 2 min read

  15. What is document parsing and why do table formats break standard text splitters?

    Beginner · 2 min read

  16. How does Parent-Document Retrieval link fine-grained vector chunks back to full parent context?

    Intermediate · 2 min read

  17. How does Reciprocal Rank Fusion (RRF) combine scores from sparse and dense retrievers?

    Intermediate · 2 min read

  18. How does contextual compression reduce context window token usage during RAG?

    Intermediate · 2 min read

  19. How does RAPTOR perform hierarchical tree-organized retrieval?

    Advanced · 2 min read

  20. How does Self-RAG train models to dynamically decide when to retrieve, evaluate, and critique documents?

    Advanced · 2 min read

Study RAG & Vector Databases with progress tracking