Archicise
Exercise

Design an AI-Powered Search Engine

Design a semantic search engine like Perplexity that understands natural language queries and synthesizes answers from multiple sources.

Functional Requirements

  • Natural language query understanding
  • Web crawling and indexing
  • Multi-source answer synthesis
  • Citation and source attribution
  • Follow-up questions and refinements
  • Real-time information retrieval

Non-Functional Requirements

  • Sub-5 second response time
  • Index billions of web pages
  • Handle 100K+ queries per second
  • Fresh index updated within hours
  • High relevance and accuracy

Questions to Consider

  • How do you combine traditional search with LLM synthesis?
  • How do you ensure freshness of information?
  • How do you handle queries requiring real-time data?
Your Solution

Query Understanding

Design the query understanding system. Consider intent classification, query rewriting, entity extraction, and handling ambiguous queries.