# Design a Data-Driven Expense Rule Evaluator Design an expense-policy evaluator whose rules are stored as validated JSON so policies can be added or changed without deploying code. ### Constraints & Assumptions - Rules use allowlisted operators and versioned schemas. - Every decision must be explainable. ### Clarifying Questions to Ask - Can rules conflict, and what precedence applies? - Who may publish a version? ### What a Strong Answer Covers - Typed rule schema, validation, safe interpretation, versioning, audit, and tests. ### Follow-up Questions - How would staged rollout work? - How would historical decisions be reproduced?