Your question is Balancing Latency and Accuracy. 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 do you balance latency and model accuracy in a real-time predictive service?
Explain how you would choose and validate candidate models, define latency objectives, and decide when a less accurate model is preferable. Address feature computation, preprocessing, model compression, caching, hardware, traffic variability, and monitoring. Provide an implementable evaluation plan that measures both predictive quality and end-to-end serving latency.