Generative AI & LLMs

Vector Databases & Embeddings

8 practice questions. Free questions open a full answer guide; the rest unlock with Pro.

  • What is an embedding, and why do we use vector similarity instead of keyword matching to find related text? Junior level
  • Why do we use a vector database with an approximate nearest-neighbor index instead of just computing the similarity to every stored vector?Go Pro Junior level
  • You're choosing the vector index for a semantic search service and the options come down to an HNSW graph versus an IVF-based index. How do you reason about that choice and tune it for your recall and latency targets?Go Pro Senior level
  • Your semantic-search service has grown to tens of millions of full-precision embeddings and the memory footprint and infra bill are getting painful. How would you reason about compressing the vectors to cut cost without quietly tanking search quality?Go Pro Senior level
  • You're picking an embedding model and dimensionality for a new semantic-search system. Walk me through how you'd choose, including the similarity metric and how you'd handle the cost of storing and searching the vectors.Go Pro Senior level
  • When you store embeddings in a vector database, how do you decide which similarity metric to use, like cosine similarity or dot product?Go Pro Junior level
  • Your vector search returns good results in a small prototype but gets slow as the collection grows into the millions of vectors. What knobs do you have to scale it, and what trade-offs do they carry?Go Pro Mid level
  • How do embeddings let a vector database do semantic search, and what determines whether two pieces of text come back as similar?Go Pro Mid level
Want questions matched to your role? Paste a job title, job description, or CV for a personalized set, or go Pro to unlock the full bank.