Design under vague distributed requirements

You are given a vague distributed-systems prompt. Show how you would clarify requirements and then design the system end to end: define the core use cases and SLAs, propose external APIs, select a data model, choose a sharding and replication strategy, justify a consistency model (strong vs eventual), and explain read/write paths. Describe failure handling (timeouts, retries, idempotency), leader election, backpressure, hot-key mitigation, caching, schema evolution, observability (metrics, logs, tracing), security, and capacity planning. Present a high-level architecture and discuss the key trade-offs you made.