Design comment ranking
Quick Overview
This question evaluates competency in designing large-scale machine learning ranking systems, testing skills such as candidate generation, label and training data strategy, feature engineering across comments/users/posts/viewer context, model architecture and serving, and operational constraints like freshness, cold start, moderation, latency, and reliability. Commonly asked in ML system design interviews, it measures the ability to balance product goals (usefulness and engagement) with constraints and trade-offs; the domain is machine learning system design and it requires both conceptual understanding and practical application.
Design a large-scale ranking system for ordering comments under a post on a community platform similar to Reddit.
When a user opens a post, the system should rank comments to maximize usefulness and engagement while demoting spam, abuse, repetitive content, and low-quality replies. Your design should cover:
- product goals and ranking objectives
- candidate generation for very large threads
- labels and training data
- feature engineering for comments, users, posts, and viewer context
- model architecture and online/offline serving
- freshness, cold start, and moderation constraints
- experimentation, metrics, and feedback loops
- latency, reliability, and fallback behavior