Your question is GenAI Developer Platform Architecture. 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).
How would you architect a GenAI developer platform to serve multiple internal teams with varying hardware and latency constraints? Discuss the platform boundary, model serving and routing strategy, resource isolation, online versus batch workloads, evaluation, observability, and failure handling. Clarify missing scale, latency, availability, compliance, and model requirements before proposing an architecture.