Your question is Low-Latency Messaging for HFT. 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 low-latency messaging system for high-frequency trading data?
Explain how you would clarify throughput, latency, ordering, durability, replay, and delivery requirements before choosing an architecture. Cover the critical online data path, failure handling, operational monitoring, and any appropriate ML components, such as anomaly detection or capacity forecasting, without placing nonessential inference on the trading-critical path.
Discuss how you would validate the design under burst traffic and partial infrastructure failure.