Design Top-K System — Section 1: Problem Understanding & Requirements

Topic: Design a Top-K system for songs / ads / videos / posts
Audience: Senior to Principal software engineers preparing for system design interviews
Output of this section: Clear scope, functional requirements, assumptions, and success criteria.

1. What “Top-K” Means

A Top-K system answers queries like:

• “Give me the top 100 trending videos in the last 24 hours.”
• “Give me the top 20 songs in the US right now.”
• “Give me the top 10 ads for a placement, filtered by policy and budget constraints.”
• “Give me the top 50 items for a user segment, based on engagement score.”

At a high level:

• We ingest events (plays, clicks, likes, purchases, skips, impressions).
• We aggregate events into metrics (views per minute, CTR, watch time).
• We compute a score for each item.
• We rank and return K items for a given scope.