NVIDIASoftware Development for LLMs

RAG & Vector Database — NCA-GENL Practice Question

A representative NVIDIA Generative AI LLMs Associate (NCA-GENL) exam question on RAG & Vector Database. Work through it below, then read why each option is right or wrong.

Short answer

The correct answer is C. FAISS, configured with an IVF index for efficient approximate nearest neighbor search.

FAISS (Facebook AI Similarity Search) is purpose-built for efficient similarity search on dense vectors and supports IVF (Inverted File) indexes that enable sub-millisecond approximate nearest neighbor queries on millions of vectors in memory. PostgreSQL B-tree indexes are not optimized for high-dimensional vector similarity search. Redis default storage does not support vector similarity natively. MongoDB's document-based queries are not designed for dense vector similarity search at this scale.

The Question

An engineering team is implementing a Retrieval-Augmented Generation (RAG) pipeline for a customer support chatbot. They need to store 10 million document embeddings and perform approximate nearest neighbor searches with sub-millisecond latency on a single server. Which vector database would be most suitable for this in-memory, single-node deployment?

APostgreSQL with standard B-tree indexes for embedding storage
BRedis with its default key-value storage engine
CFAISS, configured with an IVF index for efficient approximate nearest neighbor searchCorrect
DMongoDB with its document-based query engine

Why C is correct

FAISS (Facebook AI Similarity Search) is purpose-built for efficient similarity search on dense vectors and supports IVF (Inverted File) indexes that enable sub-millisecond approximate nearest neighbor queries on millions of vectors in memory. PostgreSQL B-tree indexes are not optimized for high-dimensional vector similarity search. Redis default storage does not support vector similarity natively. MongoDB's document-based queries are not designed for dense vector similarity search at this scale.

Why the other options are wrong

Option A: PostgreSQL with standard B-tree indexes for embedding storage

Option A does not satisfy the requirement in the scenario. Review the explanation above: the correct choice (C) is the only one that fully meets every constraint stated in the question.

Option B: Redis with its default key-value storage engine

Option B does not satisfy the requirement in the scenario. Review the explanation above: the correct choice (C) is the only one that fully meets every constraint stated in the question.

Option D: MongoDB with its document-based query engine

Option D does not satisfy the requirement in the scenario. Review the explanation above: the correct choice (C) is the only one that fully meets every constraint stated in the question.

Key idea: RAG & Vector Database

FAISS (Facebook AI Similarity Search) is purpose-built for efficient similarity search on dense vectors and supports IVF (Inverted File) indexes that enable sub-millisecond approximate nearest neighbor queries on millions of vectors in memory. PostgreSQL B-tree indexes are not optimized for high-dimensional vector similarity search. Redis default storage does not support vector similarity natively. MongoDB's document-based queries are not designed for dense vector similarity search at this scale. On the NCA-GENL exam, questions in the "Software Development for LLMs" domain test whether you can map a scenario's constraints to the right choice. Read the requirement carefully, eliminate options that violate any single constraint, and pick the one that satisfies all of them with the least operational overhead.

Ready to see how you'd score?

Take the free practice quiz and find out which NVIDIA Generative AI LLMs Associate domains you need to focus on. No signup required.

Practice 5 similar questions

Same cert, same or adjacent domain. Use these after reviewing the explanation.

Related NVIDIA Generative AI LLMs Associate Practice Questions