Your question is Handling Peak Traffic Spikes. 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 architect a system to handle massive traffic spikes during peak auction hours?
Discuss the ML-enabled request path for Whatnot auction discovery and bidding-related recommendations, including capacity planning, graceful degradation, and operational safeguards. Cover retrieval, ranking, online and batch components, evaluation, and failure handling without assuming that every model remains available during a spike.