Design client-side event batching and prevent unload data loss

It tests designing a high-throughput telemetry pipeline that minimizes network overhead without dropping events during page transitions. A strong answer outlines an in-memory ring buffer with size and time triggers, flushes via fetch during the session, and escalates to sendBeacon or fetch keepalive on visibilitychange or beforeunload to survive page death. It should also mention an IndexedDB retry queue for offline or failed requests.