Whenever you're ready, walk me through how you'd approach this.
Walk me through how you would build a data pipeline that must handle sudden, unpredictable spikes in incoming streaming data volume for McKinsey & clients. Explain your ingestion, buffering, processing, storage, and serving choices, including how the design scales without data loss or unacceptable latency. Cover backpressure, autoscaling, ordering, duplicate handling, failure recovery, observability, capacity planning, and cost controls.