14Your VP asks whether we should use an open-weight model instead of a commercial API. Lay out the trade-offs.▼hardNewMistral AIMetaDatabricks◆ premiumA trade-off question where both sides have serious arguments. The interviewer wants the dimensions that decide it, and a recommendation that depends on facts you would go and get.Practise this question →Read answer →
22You're launching a new AI product for customers in the US, EU and India. How do you choose the model stack?▼expertNewMicrosoftAtlassianDatabricks◆ premiumThere is no single best model, and the choice will change within a year. Principal answers design a selection process and an architecture that make the next choice cheap.Practise this question →Read answer →
09Design an AI writing assistant for our B2B product. Now legal says EU customers' data must never leave the EU. What changes?▼hardNewAtlassianNotionCanva○ sign inA design question with a constraint dropped halfway. The interviewer wants to see residency flow through every component: model calls, logs, caches, embeddings, evals and support access.Practise this question →Read answer →
21A national government wants your models deployed entirely within its borders and under its control. How do you approach it?▼expertNewMicrosoftGoogleMistral AI◆ premiumSovereign deployments combine product, infrastructure, security, policy and commercial questions. Principal answers separate what the customer needs (control, residency, continuity, local language) from what they ask for, and design a deployment model the company can sustain.Practise this question →Read answer →
29Sales wants EU and India data residency included free for enterprise API customers. Finance wants a 10% premium. Decide.▼hardNewOpenAIAnthropicGoogle◆ premiumRegional capacity is fragmented capacity, and four providers charge about 10% for it. Price it, define exactly what you promise, and tell customers which models arrive late in their region.Practise this question →Read answer →