Design a content moderation platform
Company: ByteDance
Role: Software Engineer
Category: ML System Design
Difficulty: medium
Interview Round: Technical Screen
Design a large-scale content moderation system for a short-video platform.
Users can upload videos, captions, comments, audio, and other metadata. The system should detect policy violations such as spam, nudity, violence, hate speech, self-harm, and copyright abuse. It should support both pre-publication checks and post-publication monitoring.
Discuss:
- Functional requirements and moderation outcomes
- Online and asynchronous processing pipelines
- Multimodal ML inference for text, image, video, and audio
- Rules engine, risk scoring, and policy decisions
- Human review workflows, escalation, and appeals
- Model training, feedback loops, and ML infrastructure
- Latency, throughput, reliability, and regional compliance
- Metrics for model quality and operational effectiveness
Overview: This question evaluates system design and machine learning engineering competencies within the ML System Design domain for building large-scale, multimodal content moderation platforms, focusing on skills such as scalable inference, pipeline architecture, policy enforcement, human review workflows, compliance, and operational metrics.