Your question is Design a Production Rollback Plan. Take a moment with it on the right.
Talk me through your thinking if you like. When you're confident, submit your answer and I'll grade it like a real screen (7/10 or better passes).
You are preparing a production deployment for a critical backend service, and you want a clear plan in case the release causes errors, latency spikes, or downstream failures after traffic is shifted. The goal is not just to undo the change, but to restore service safely, minimize rider and driver impact, and avoid making the incident worse during a high-pressure rollout.
How would you design a rollback plan for a failed production deployment?