Your question is Balancing Quality and Compute Costs. 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 model performance (accuracy/quality) with computational costs?
Discuss how you would make this tradeoff when designing and operating an ML system. Cover model selection, inference architecture, latency and infrastructure constraints, offline and online evaluation, and how you would monitor quality, cost, and regressions. Be explicit about when additional computation is justified and when a simpler model or staged architecture is preferable.