System Design Interview: Design YouTube w/ a Ex-Meta Staff Engineer

Designing a scalable video-sharing platform requires balancing high availability with eventual consistency while managing massive data volumes. The system architecture centers on asynchronous processing, where raw video uploads are handled via multi-part requests to bypass gateway limitations. To ensure low-latency streaming across diverse network conditions, videos undergo transcoding into multiple resolutions, enabling adaptive bitrate playback. This process relies on chunking content into small, two-second segments, which are then served through a Content Delivery Network (CDN) using manifest files to optimize delivery. By decoupling metadata storage from blob storage and employing stateless services, the system achieves horizontal scalability, capable of supporting millions of daily uploads and views. This approach prioritizes a seamless user experience, allowing for immediate playback initiation even when network bandwidth fluctuates significantly.