Good question! The queries go directly to Twitter's API (via a third-party provider).

  No local index or stored tweets—everything is fetched in real-time when you search or view a profile. We do cache responses temporarily to avoid hitting rate limits, but the cache expires quickly.

  The benefit of this approach:
  • Always up-to-date (real-time data)
  • No storage costs or maintenance
  • Simpler architecture

  The downside:
  • Dependent on API availability
  • Rate limit considerations

  I considered building a local index but decided against it for now—wanted to ship fast and keep it simple. Might revisit if the API approach becomes problematic.