A Dependency Having a Bad Day
Decide what you would build from the brief alone. Everything below it is available, but the exercise stops working if you open it first.
Your payment provider starts returning intermittent 503s and slow responses. Your service currently has no timeouts and no retries. Design the policy.
Adding three retries with a short delay, because the errors are intermittent and retrying makes the error rate drop in staging. Against a provider that is failing because it is overloaded, this triples your request rate at its worst moment — and the timeouts that were already happening now cost three timeouts each, so your own workers stay occupied three times as long. The provider's recovery is delayed by the traffic, and your service saturates before it does.