Architecture tradeoffs, scaling strategies, and reliability at senior level.
I begin by clarifying scope, traffic assumptions, and non-functional constraints, then draft a simple baseline architecture before scaling it step by step. This shows I can reason from first principles rather than jumping to complex components too early.
I align the decision with business impact of stale data versus failed writes during partitions. Critical financial operations may prefer stronger consistency, while social feeds often prioritize availability with eventual consistency.
I would combine caching tiers, read replicas, graceful degradation paths, and strong observability with SLO-based alerting. For resilience, I include rate limiting, backpressure, and controlled failover tested via game-day scenarios.