Your question is Batch vs Real-Time Inference Trade-offs. 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).
Discuss the trade-offs between batch processing and real-time inference for a personalized recommendation engine at Intuit.
Explain when recommendations should be precomputed, generated online, or served through a hybrid architecture. Address freshness, personalization quality, latency, infrastructure cost, feature availability, privacy, failure handling, and offline and online evaluation. Include a recommendation for an initial design and describe how you would evolve it as usage and personalization requirements grow.
