Fintech Engineering Handbook
Points and comments are a snapshot, not live.
A handbook on software engineering patterns for systems where money is the primary focus.
The handbook covers three core principles: no invented data (enforced via idempotency, deduplication, reconciliation), no lost data (via full precision, at-least-once delivery, event sourcing, audit trails, immutability), and no trust (verifying webhooks, cross-checking data, failing loudly).
It details money representation (floating-point, arbitrary precision, minor-units integer, rational numbers), rounding strategies, currency handling, FX rates, and the ledger-double-entry bookkeeping, value/booking/settlement time, audit trails, event sourcing, immutability, and reversals/corrections.
What commenters are saying
Commenters largely validated the handbook as accurate, drawing on their own fintech experience. The top-rated comment touted idempotency keys as the biggest lesson. A debate emerged on representing monetary amounts: some advocated for integer minor-units, others warned it fails with partners having different decimal places, and recommended strings or explicit mantissa/exponent pairs. One commenter recommended understanding accounting principles beyond just debits/credits.