Whenever you're ready, walk me through how you'd approach this.
Walk me through your approach to capacity planning for an enterprise platform experiencing unpredictable, seasonal traffic spikes.
Treat this as an ML platform design discussion. Explain how you would gather missing workload data, forecast demand, size online inference and supporting services, and validate the plan. Discuss autoscaling, quota management, cost controls, resilience, and how you would monitor both infrastructure and model-serving behavior. Be explicit about assumptions, uncertainty, and how you would handle a spike that exceeds forecast capacity.