I helped redesign a large BigQuery-based enterprise data warehouse and cut spend by 57%. The biggest savings didn't come from dashboards or one-off query tuning. They came from architecture decisions — partitioning, clustering, incremental MERGE patterns, and a better capacity model. Here's how I approach cost as an architectural problem in large systems. The Problem I joined as the Solution Architect for a Fortune 500 financial company running BigQuery at scale — hundreds of analysts, dozens of automated pipelines, 4+ TB of daily ingestion across finance, risk, and regulatory reporting workloads. The platform had grown organically. Datasets were added without architectural standards. Queries ran without cost awareness. The pricing model was never revisited as workloads matured.…