Your question is Low-Latency Inference Service. 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 design an inference service that meets low-latency requirements for real-time document analysis for Docusign?