Fraud Detection API — FastAPI Inference Service

This project provides a production-style machine learning inference API for detecting fraudulent credit card transactions.

The API serves a trained XGBoost fraud detection model via FastAPI, with strict input validation, feature alignment, and safe inference handling.

The goal of this project is to demonstrate how a trained ML model is exposed, validated, and consumed in a real-world system — not just trained.

Problem Context

Credit card fraud detection is a highly imbalanced classification problem, where missing a fraudulent transaction is often more costly than flagging a legitimate one.

In real production systems, models are rarely used directly. They must be:

• Validated
• Properly preprocessed
• Safely deployed behind an API

This project focuses on the serving and inference layer of a fraud detection system.

Key Features

• FastAPI-based inference service
• XGBoost fraud detection model
• Strict request validation using Pydantic schemas
• Feature name and order alignment with training schema
• Partial preprocessing during inference (Time & Amount scaling)
• Clear error handling with meaningful HTTP responses
• Model metadata endpoint for observability
• JSON-safe prediction outputs

Project Structure

fraud-detection-api/
├── app/
│   ├── main.py        # API endpoints
│   ├── schemas.py     # Request validation schemas
│   ├── inference.py  # Model inference logic
│   ├── config.py     # Centralized configuration
│   └── __init__.py
├── models/
│   └── xgboost.pkl
├── artifacts/
│   └── standard_scaler.pkl
└── .gitignore

---

📌 API Endpoints

### Health Check :
Returns API health status.

### Model Information :
Returns model metadata such as:
- Model name
- Version
- Threshold
- Number of features

### Fraud Prediction
Accepts a full feature vector (Time, Amount, V1–V28) and returns
a fraud probability and decision.

Example Prediction Request

{
  "Time": 50000,
  "Amount": 120.75,
  "V1": 0.12,
  "V2": -0.45,
  "V3": 0.33,
  "V4": -0.08,
  "V5": 0.21,
  "V6": -0.15,
  "V7": 0.04,
  "V8": -0.02,
  "V9": 0.11,
  "V10": -0.19,
  "V11": 0.07,
  "V12": -0.13,
  "V13": 0.00,
  "V14": -0.22,
  "V15": 0.09,
  "V16": -0.17,
  "V17": 0.05,
  "V18": -0.06,
  "V19": 0.18,
  "V20": -0.03,
  "V21": 0.14,
  "V22": -0.20,
  "V23": 0.02,
  "V24": -0.09,
  "V25": 0.00,
  "V26": -0.11,
  "V27": 0.08,
  "V28": -0.04
}

---

📌 Design Decisions

- The API expects the **same feature schema used during training**
  to avoid inference drift.
- PCA features (V1–V28) are assumed to be computed upstream.
- Only Time and Amount are scaled during inference.
- Feature order is explicitly enforced before prediction.
- Inference logic is separated from API routing and configuration.

Tech Stack

• Python
• FastAPI
• Pydantic
• XGBoost
• Scikit-learn
• Pandas

---

Author

Mohamed Saad