Design a Top-K trending items service

Question

Design a system that continuously identifies the Top-K trending items on a large social platform (for example, hashtags, videos, or posts).

Millions of user interactions (views, likes, shares, comments) are generated every second. The system should process these events in near real-time and continuously update the Top-K trending items over a sliding time window (e.g., the last 10 minutes, 1 hour, or 24 hours).

Your design should address:

Real-time ingestion of high-volume user events

Aggregation and ranking of items by popularity

Handling large-scale distributed data

Trade-offs between exact vs approximate Top-K computation

Fault tolerance and scalability

Serving results with low latency

Assume the system must support billions of events per day and serve trending lists to millions of users globally.

Accepted Answer

A detailed reference solution is available to Crack ML Interview members.