FastAPI Model Gateway

This project implements an API Gateway using FastAPI that routes incoming requests to different ML models or processing endpoints.
It also includes query routing, preprocessing, and model inference logic for modular and scalable design.

---

🚀 Features

• FastAPI-powered REST API
• Query Routing Layer for directing requests to appropriate models
• IsolationForest-based Anomaly Detection
• Joblib-based Model Serialization
• Dynamic Column Transformation using ColumnTransformer
• Pydantic Models for request validation
• Multithreaded Request Handling
• Endpoints with Form, File, and JSON support
• Modular structure with document_router for endpoint documentation

---

🧩 Project Structure

API_GATEWAY/
│
├── main.py                  # Entry point (FastAPI app)
├── query_routing.py         # Query routing logic
├── enpoint_documentation.py # Endpoint descriptions
├── model.joblib             # Serialized ML model
├── requirements.txt         # Dependencies
└── README.md                # Project documentation (this file)

---

⚙️ Installation

1️⃣ Clone the Repository

git clone https://github.com/sarthakm402/API_GATEWAY.git
cd api-gateway

2️⃣ Create and Activate a Virtual Environment

python -m venv venv
venv\Scripts\activate   # On Windows
source venv/bin/activate  # On Linux/Mac

3️⃣ Install Dependencies

pip install -r requirements.txt

---

🧠 Model Setup

The project uses a serialized Isolation Forest model trained for anomaly detection.
You can replace model.joblib with any scikit-learn compatible model.

To retrain and save a model:

from sklearn.ensemble import IsolationForest
from joblib import dump

model = IsolationForest()
model.fit(X_train)
dump(model, "model.joblib")

---

🚪 Running the Server

To start the FastAPI app:

uvicorn main:app --reload

Server runs on:

http://127.0.0.1:8000

You can also access:

Swagger UI: http://127.0.0.1:8000/docs

---

📡 Example Endpoints

➤ /predict — Model Inference

Accepts JSON, CSV, or form data.
Runs preprocessing, routes query, and returns prediction.

Example Request:

curl -X POST "http://127.0.0.1:8000/predict" -H "Content-Type: application/json" -d '{
  "feature1": 3.4,
  "feature2": "A",
  "feature3": 12
}'

Example Response:

{
  "prediction": "normal"
}

➤ /upload — Upload File Endpoint

Uploads CSV files for batch prediction.

➤ /docs — API Documentation

Interactive documentation powered by Swagger UI.

---

🧠 Query Routing

The query_routing.py file defines rules to direct incoming data to the appropriate model or processing function.
For instance:

if "anomaly" in query_type:
    route_to(isolation_forest_model)
else:
    route_to(default_model)

---

🧰 Environment Variables

If you want to use environment-specific configurations, create a .env file:

MODEL_PATH=model.joblib
PORT=8000
LOG_LEVEL=info

---

🧪 Testing

Run tests using:

pytest

---

📜 License

This project is licensed under the MIT License — see the LICENSE file for details.

---

👨‍💻 Author

Sarthak Mohapatra
💼 AI Developer | ⚙️ Machine Learning | 🧠 FastAPI & Model Deployment