Design real-time top-k items
Company: Amazon
Role: Software Engineer
Category: System Design
Difficulty: hard
Interview Round: Technical Screen
Design a system to compute and serve the real-time top-k most purchased items from a continuous stream of purchase events records(customer_id, item_id, timestamp). Support:
(a) top-k over a sliding time window T (e.g., last 1 hour) and
(b) daily rolling totals. Discuss ingestion, partitioning, exactly-once/at-least-once semantics, handling late or out-of-order events, data structures for efficient updates (e.g., count map + min-heap or approximate sketches), storage choices, API/query layer, scalability and fault tolerance, and latency vs. accuracy trade-offs.
Quick Answer: This question evaluates understanding of real-time streaming system architecture and stateful analytics, covering event-time processing and late/out-of-order handling, ingestion and partitioning strategies, state management and checkpointing, fault tolerance, and algorithms for top-k aggregation.
