Banking synchronization
Plaid transaction sync uses encrypted access tokens, retry behavior and idempotent processing to handle duplicate and partial events safely.
Python fintech backend
PayControl
A complete banking and spending-policy backend built from the ground up.
PayControl gives parents real-time visibility and policy control over family spending. Its backend covers authentication, banking synchronization, transaction processing, rule evaluation, live events, reporting and the operational safeguards required around financial data.
PayControl · Policy engine
PLAID WEBHOOK
COFFEE CO
$6.40
ALLOWPOLICY ENGINE
DAILY LIMIT✓
CATEGORY✓
MERCHANT✓
VELOCITY✓
ALLOW
12
NOTIFY
4
REVIEW
2
Live simulation of the system's real behavior. Private product repository. Architecture facts are derived from the Python API, service layer, database migrations and automated backend test suite.
Operating loop
01Plaid webhooks
02Idempotent sync
03Transaction ledger
04Rules evaluation
05Live notifications
06Reports & audit
50
FastAPI endpoints
36
automated tests
6
Alembic migrations
24/7
event-driven backend
System anatomy
Complexity broken into reliable boundaries.
Policy engine
Configurable spending rules evaluate transaction behavior and produce violations and notifications without coupling policy logic to the client apps.
Live operational events
Server-sent events and scheduled processing keep clients current while preserving a clear backend source of truth.
Auditable reporting
PDF and structured JSON reports provide exportable outcomes over a migration-backed PostgreSQL domain model.
Engineering judgment
The decisions behind the build.
Seniority is not the number of technologies in a project. It is the ability to identify failure modes early and choose boundaries that keep the system understandable after launch.
Financial events must be idempotent
Retries and duplicate webhooks cannot create duplicate transactions or violations, so sync state is explicit and recoverable.
Secrets stay encrypted
Plaid access tokens and sensitive configuration are isolated behind the backend rather than exposed to product clients.
Policy belongs in the domain layer
Spending rules are modeled as testable backend behavior so iOS and future clients consume the same decisions.
Production stack
Chosen for the system.
PythonFastAPIPostgreSQLSQLAlchemyAlembicPlaidAPSchedulerSSEDockerRenderPytest