Flowlet Code

This directory contains the core application code for Flowlet, an embedded finance platform that exposes payments, wallets, card issuance, and compliance through a unified API. The code is organized into two primary areas: the Python Flask backend services and the machine learning services that power AI-enhanced financial intelligence.

Directory Structure

code/
├── backend/           # Core Python microservices and API layer
└── ml_services/       # Machine learning models and AI services

Backend

The backend/ directory houses the main application server built with Python and Flask. It follows a modular architecture organized by domain, with clear separation between routes, models, services, and integrations.

Key Components

Directory	Purpose
app.py	Application entry point with the create_app() factory function
src/routes/	Flask Blueprints defining REST API endpoints for auth, payments, wallets, cards, compliance, and analytics
src/models/	SQLAlchemy ORM models for users, accounts, transactions, cards, and ledger entries
src/services/	Core business logic for payments, cards, wallets, and account management
src/compliance/	KYC/AML engines, regulatory frameworks, and compliance workflows
src/security/	Authentication, encryption, rate limiting, threat prevention, and password security
src/integrations/	Connectors for external systems including banking (Plaid, FDX), payments (Stripe), and currency exchange
src/analytics/	Dashboard services, reporting engines, real-time analytics, and metrics calculation
src/nocode/	Rule engine, workflow builder, and configuration engine for dynamic business logic
src/gateway/	API gateway logic for request routing and external API handling
src/config/	Centralized application settings and security policies
src/database/	Database utilities and connection management
src/utils/	Shared utility functions and helpers
migrations/	Database schema migration files
tests/	Comprehensive test suite covering unit, integration, functional, API, performance, and security testing

Running the Backend

Development server:

cd backend
python run_server.py

With Docker:

cd backend
docker-compose up

Running tests:

cd backend
./run_tests.sh

For full backend documentation, see backend/README.md.

ML Services

The ml_services/ directory contains the machine learning stack that powers real-time fraud detection, risk assessment, and intelligent financial insights. Models are built with scikit-learn, XGBoost, and LightGBM, exposed as services that the backend can call during transaction processing.

Key Components

Directory	Purpose
ai_models/	Core AI model implementations including fraud detection, risk assessment, transaction intelligence, and support chatbot
fraud_detection/	Production fraud detection pipeline with ensemble stacking, anomaly detection, and supervised models
tests/	ML-specific tests for model correctness, fairness, and performance benchmarks

Fraud Detection Ensemble

The fraud detection system uses a stacked ensemble combining XGBoost, LightGBM, and Random Forest models. This architecture achieves strong performance on production transaction data.

Metric	Value
AUC-ROC	0.987
Precision	97.3%
Recall	96.1%
F1 Score	96.7%
Inference latency (p50)	8 ms

Running ML Services

ML services are typically invoked by the backend or run as standalone inference workers. Individual models can be tested or trained by running their respective module files directly. For operational deployment, the fraud detection service in fraud_detection/service.py provides the main entry point for real-time scoring requests.

Technology Stack

Layer	Technology
Backend language	Python 3.11+
Web framework	Flask + flask-restx
ORM	SQLAlchemy
Databases	PostgreSQL, Redis
Messaging	Kafka / RabbitMQ
ML frameworks	scikit-learn, XGBoost, LightGBM
Containerization	Docker
Testing	pytest, with unit, integration, functional, performance, and security suites

Testing

Both backend and ML services maintain extensive test coverage. The backend test suite is organized by test type under backend/tests/, while ML service tests live under ml_services/tests/. Tests validate correctness, integration contracts, performance benchmarks, and security controls.

Related Documentation

• Backend README - Detailed backend architecture, API routes, and configuration
• Project README - High-level project overview, installation guide, and contribution guidelines
• docs/ - Additional documentation including API reference, architecture specs, and ML performance reports