Design Semantic Document Search · System Design Interview

Design Semantic Document SearchDesign a search service that finds documents by meaning rather than by exact words, over a large corpus that keeps changing.What it has to do

Return the documents closest in meaning to a natural-language query

Make a new or changed document searchable shortly after it is written

Combine meaning with filters such as owner, date and language

Scale

Documents: 2 billion

Queries per second at peak: 5,000

Out of scope

Training the embedding model

Answer generation on top of the results

Crawling the sources

Access control inside the documents themselves

Practise this challengeAll system design challenges

LetsDiscuss

Loading your session…