🛡️ FinSec Shield: Real-Time Credit Card Fraud Detection

An end-to-end, production-ready Machine Learning microservice that detects fraudulent credit card transactions in real-time. This project encompasses the entire ML lifecycle: from exploratory data analysis and handling severe class imbalance, to model training, footprint optimization, and deploying a distributed cloud architecture.

🔗 Live Interactive Dashboard (Note: The AI backend is hosted on Render's free tier. If the server has been inactive, the very first inference may take 30-50 seconds to wake up. Subsequent inferences execute in <50ms).

---

🏗️ System Architecture (Microservice)

To replicate enterprise-grade deployments, this system is split into two distinct environments:

1. The Brain (Backend API on Render): A robust FastAPI server hosting a highly optimized 31 KB LightGBM model. It handles the heavy tensor math, feature validation via Pydantic, and returns a strict JSON decision payload.
2. The Face (Frontend UI on Vercel): A zero-dependency, vanilla HTML/CSS/JS enterprise dashboard. Hosted on Vercel's edge network for lightning-fast global delivery, it securely routes transaction payloads to the Render API.

---

🧠 Machine Learning Engineering

• Dataset: Highly imbalanced credit card transaction dataset (fraud accounts for <0.2% of data).
• Feature Engineering: Processed 30 features (Time, Amount, and PCA-transformed variables V1-V28).
• Leakage Prevention: Strictly isolated training and validation sets to prevent data leakage during scaling and sampling techniques.
• Model Optimization: Trained a Gradient Boosted Tree (LightGBM) optimized specifically for inference speed and high Recall to minimize false negatives in fraud detection.

---

💻 Tech Stack

• Machine Learning: LightGBM, Pandas, Scikit-Learn, Joblib
• Backend: Python, FastAPI, Uvicorn, Pydantic
• Frontend: HTML5, CSS3, Vanilla JavaScript (Fetch API)
• Deployment: Vercel (Frontend), Render (Backend API)

---

🚀 Local Installation & Setup

Want to run this system on your local machine?

1. Clone the repository

git clone https://github.com/nilotpaldhar2004/Credit-Card-Fraud-Detection.git
cd Credit-Card-Fraud-Detection

2. Create virtual environment & install dependencies

python -m venv venv

Activate environment:

Mac/Linux:

source venv/bin/activate

Windows:

venv\Scripts\activate

Install dependencies:

pip install -r requirements.txt

3. Run backend server

python app.py

4. Open dashboard

Go to:

http://127.0.0.1:8500

---

📡 API Reference

POST /predict

Evaluates a 30-feature transaction for fraud detection.

Request (JSON)

{
  "Time": 1205.0,
  "V1": -0.8, "V2": 0.5, "V3": 1.2, "V4": 0.1, "V5": -0.3,
  "V6": 0.0, "V7": 0.8, "V8": -0.1, "V9": 0.4, "V10": -0.2,
  "V11": 0.5, "V12": 0.1, "V13": -0.5, "V14": -1.2, "V15": 0.3,
  "V16": 0.4, "V17": 0.8, "V18": -0.2, "V19": 0.5, "V20": 0.1,
  "V21": 0.0, "V22": 0.1, "V23": -0.2, "V24": 0.3, "V25": 0.1,
  "V26": -0.1, "V27": 0.0, "V28": 0.0,
  "Amount": 250.00
}

Response (JSON)

{
  "status": "success",
  "fraud_probability": 0.0034,
  "is_fraud": false,
  "action": "APPROVE",
  "latency_ms": 42.1
}

---

👨‍💻 Author

Nilotpal Dhar
Computer Science Student (AOT)
Specializing in Machine Learning, Model Deployment, and API Development