YouTube data platform — Start Here
Scope
Signals in
Products out
Interview Script
“I’d design YouTube’s analytics and data platform. Playback, engagement, search, impression, ad, and backend events enter a durable Kafka backbone. Streaming jobs serve freshness-sensitive products such as live views, trending, QoE alerts, and online features. The same events land in an Iceberg lakehouse, where batch jobs validate, deduplicate, reconcile late data, and publish certified Silver and Gold products for Creator Studio, recommendations, experiments, monetization, BI, and backfills—with governance and quality around every layer.”
Capabilities
Measure
Collect and qualify audience behavior
Decide
Turn fresh signals into product decisions
Operate
Serve creators, revenue, and trust
Platform Flow
YouTube signals
Fast path
seconds → minutesTruth path
hourly → T+1“Playback and business events enter one replayable platform. Streaming serves freshness-sensitive products, while Iceberg and batch jobs publish certified data for creators, recommendations, monetization, experiments, and backfills.”
Scale Targets
Interview assumptions for sizing and trade-off discussion.