Your question is Low-Latency Trading Component. 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 a high-frequency trading system component to minimize latency?
Treat this as an end-to-end ML system design problem. Explain how you would clarify the component's responsibilities, separate offline model development from the online execution path, choose models and serving infrastructure, and protect correctness when dependencies or models fail. Discuss how you would measure latency, throughput, model quality, feature freshness, and training-serving consistency.