Overview

I’m building a production-style banking core focused on correctness and auditability. The system models customers, accounts, and a double-entry ledger that records every money movement as balanced debits and credits. Transfers are processed transactionally with idempotency keys to prevent duplicates, and all changes are traceable via immutable audit logs.

Architecture

• Services: account-service (open/close/list), transfer-service (initiate/commit), reporting-service (balances/statements).
• Ledger: Double-entry model: each transfer posts two LedgerEntry rows (debit/credit) that must balance to zero.
• Persistence: PostgreSQL, JPA/Hibernate; Flyway for schema migrations; optimistic locking on balances.
• Security: Spring Security + JWT (roles: Customer, Operator, Admin); request idempotency via Idempotency-Key header.
• APIs: REST with OpenAPI/Swagger; example endpoints: POST /api/accounts, POST /api/transfers, GET /api/accounts/{id}/balance, GET /api/statements?month=....
• Messaging (optional): Kafka events (TransferCreated, LedgerPosted) to drive async notifications or downstream analytics.
• Deployment: Docker Compose (app + Postgres + Kafka optional); profiles for dev and prod.

Highlights

What I Learned

• Designing financial domains with clear invariants and isolation levels that avoid race conditions.
• How to structure Spring apps with controllers → services → repositories and keep business logic testable.
• Using Testcontainers to run integration tests against a real Postgres instance in CI.

Next Steps

• Outgoing webhooks and email notifications for posted transfers.
• Limits & fraud rules (velocity checks, daily caps, blacklist/whitelist).
• Batch reconciliation job and downloadable PDF statements.