10 parts · 18 chapters

Ledgers, Payments and Reconciliation

The backend mirror of Money, Identity and Trust. A ledger is the one system in a fintech that must never be wrong, and the techniques that keep it right are old: double-entry bookkeeping, append-only journals, idempotency and reconciliation. This course teaches them as code.

Ten parts: double-entry accounting for engineers; ledger schemas, accounts, balances and holds; idempotency end to end; payment state machines; rails, webhooks and callbacks; reconciliation and breaks; settlement, float and FX; fees, interest and rounding; audit, regulators and reporting; and a capstone ledger service with a reconciliation job.

double entry · ledger schemas · idempotency · payment state machines · rails and webhooks · reconciliation · settlement and FX · fees, interest and rounding · audit and regulators · capstonesenior → staff · engineers building wallets, lending, payments or banking backends
double entryDebits, credits, account types and the invariant that every posting sums to zero.
schemasAccounts, journal entries, postings, balances, holds and versions.
idempotencyKeys, request fingerprints, stored responses, and idempotent consumers.
state machinesTransfers, payouts and card authorisations as explicit states.
reconciliationMatching the ledger with banks and processors, and working breaks.
money mathsMinor units, rounding rules, interest accrual, FX and fees.
Built on Trust and ScalingAssumes Money, Identity and Trust (the client side of money flows), SQL and PostgreSQL Internals, and Backend System Design part 3.