Question
A fintech ingests transaction feeds from ~30 partner banks daily (each its own file format and delivery cadence) into a ledger that must reconcile to the penny against each bank's own control totals. Files sometimes arrive late, partially, duplicated, with a changed column, or with totals that don't add up — and a wrong ledger is a compliance and money problem. Design a data-quality and reconciliation layer that decides, per feed, whether to accept, quarantine, or reject the day's data, and that catches partial/duplicate/changed deliveries before they corrupt the ledger.
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.