
You’ve been asked to lead an engineering effort to handle a new workload in which a platform must ingest a high volume of API calls, transform the inputs, and produce downstream results under tight performance constraints. The use case is important because delayed or dropped processing directly affects customer-facing workflows, but the right architecture is still unsettled: some engineers want synchronous request-time processing, others want an asynchronous pipeline, and the infrastructure team is concerned about cost and operational complexity.
Given a series of API calls or inputs, how would you design and execute a solution to process the data within tight performance constraints?