Scale a rules engine for high traffic

## Follow-up: scaling Assume the expense rules evaluator will be used in production and must support large-scale access. Design a high-level system that can: - Evaluate rules for new expenses in near real-time (e.g., at authorization time or shortly after purchase) - Support frequent reads (e.g., dashboards showing violations) - Allow admins to create/update rules via an API - Remain reliable under high QPS and large volumes of expenses Discuss: - architecture and components - data storage - caching/compilation of rules - consistency/latency tradeoffs - observability and failure handling