Your question is Automate Zero-Downtime Service Deployment. 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 need to automate deployment of a new production service that will sit on the critical path for API traffic. The service must be released through a CI/CD pipeline with no visible downtime, and you cannot rely on manual approval steps during the cutover except for an emergency stop. The challenge is not just shipping the first version, but making repeat deployments safe as traffic scales, dependencies change, and schema updates are introduced. You also need confidence that a bad release can be detected quickly and rolled back without creating a longer outage than the deployment itself.
How would you automate the deployment of a new service with zero-downtime requirements?