Question
Design a reverse-ETL system that syncs computed audiences and customer attributes FROM the warehouse back OUT to operational SaaS tools (Salesforce, a CRM, ad platforms, an email tool). Marketers define an audience as a SQL query over the warehouse (e.g. 'high-LTV churn-risk users'); the system must keep each destination in sync as the query's result set changes daily, push only the diffs (not re-send millions of unchanged rows), respect destination API rate limits, and never push a user who has opted out. ~50M warehouse rows, dozens of syncs, destination APIs that are slow and rate-limited.
Clarify scale and constraints first. Propose a clean component breakdown, then go deep on the hard parts — data model, bottlenecks, consistency, failure modes — and name the trade-offs you are making.