Apppublic

executor1389/RealTimeCreditCardFraudDetectionApi

sourceHugging Faceupdated 7mo agoView on Hugging Face

0likes

App README

FraudShield AI: Real-Time Credit Card Fraud Detection

Website: https://huggingface.co/spaces/executor1389/RealTimeCreditCardFraudDetectionApi

[image]

![Fraud Detection CI/CD](https://github.com/dpetrovic89/RealTimeCreditCardFraudDetectionApi/actions)

FraudShield AI is a production-ready MLOps ecosystem designed to detect fraudulent credit card transactions in real-time. Built with a 100% free-tier stack, this project demonstrates senior-level engineering across the entire data-to-deployment lifecycle.

🏗️ Architecture Stack

FraudShield AI implements a layered architecture from data ingestion to real-time serving. See the **System Architecture Guide** for full details and diagrams.

—Model: LightGBM (89% Recall on real Kaggle data)
—Serving: FastAPI + Docker (serving on port 7860/8000)
—Frontend: Premium Glassmorphic UI
—Monitoring: Evidently AI + Prometheus
—CI/CD: GitHub Actions (Auto-Sync to Hugging Face)

📊 Data Privacy & Source

—Real Data: The project is designed for the Kaggle Credit Card Fraud Detection dataset. Due to its 150MB size, it is git-ignored and stored locally in data/creditcard.csv.
—Mock Data: For CI/CD and testing on the cloud, a scripts/generate_data.py utility creates a synthetic dataset that mimics the real distribution without exposing sensitive information.
—Model Storage: The "wisdom" of the real data is compressed into the 7KB src/model.txt used by the API.

🚀 Step-by-Step Deployment Guide

Follow these steps to get the full stack running on your machine.

1. Prerequisites

Ensure you have the following installed:

—Python 3.10+
—Docker Desktop
—Minikube (for Kubernetes deployment)
—Git

2. Local Setup

bash

# Clone the repository
git clone https://github.com/dpetrovic89/RealTimeCreditCardFraudDetectionApi.git
cd RealTimeCreditCardFraudDetectionApi

# Install dependencies
pip install -r requirements.txt

3. Run with Python (Development)

bash

# Start the FastAPI server
uvicorn src.app:app --host 0.0.0.0 --port 8001 --reload

Access the frontend by opening frontend/index.html in your browser.

4. Run with Docker

bash

# Build the image
docker build -t fraud-detection-api .

# Run the container
docker run -p 8001:8000 fraud-detection-api

5. Deploy to Kubernetes (Minikube)

bash

# Start Minikube
minikube start

# Apply manifests
kubectl apply -f k8s/deployment.yaml

# Expose the service
minikube service fraud-detection-service

6. Deploy to Hugging Face Spaces (ML Optimized & Free)

1.Create a "New Space" on Hugging Face.
2.Select Docker as the SDK.
3.Push your GitHub repository (or upload files).
4.Hugging Face will build the Dockerfile and host your app on port 7860!

7. Deploy to Vercel (Fastest & Free)

This app is optimized for Vercel Serverless Functions.

1.Install Vercel CLI: npm i -g vercel
2.Run vercel in the root directory.
3.Your frontend will be at https://your-project.vercel.app.

8. Activate CI/CD

To enable automated builds and tests:

1.Go to your GitHub Repo > Settings > Secrets and variables > Actions.
2.Add the following secrets:
3.DOCKERHUB_USERNAME: Your Docker Hub username.
4.DOCKERHUB_TOKEN: Your Docker Hub Access Token.
5.Every push to main will now trigger a build and push to Docker Hub!

📊 Monitoring & Data Quality

—Validation: Run python scripts/data_validation.py to check data health via Great Expectations.
—Drift: Run python src/monitor.py to generate an Evidently AI report (monitoring/drift_report.html).
—Metrics: Prometheus scrapes API performance data on port 8001.

📄 License

This project uses the Kaggle Credit Card Fraud Detection dataset. Distributed under the MIT License.