FastAPI ML Project — Car Price Prediction API

A production-style FastAPI boilerplate for serving Machine Learning models, built around a car price prediction use case. Includes Redis caching, Prometheus metrics, and Grafana dashboards out of the box.

Features

• ⚡ FastAPI application with a clean, modular project structure
• 🤖 ML model serving via joblib (car price prediction)
• 🗄️ Redis caching for prediction results, keyed on input features
• 🔐 JWT-based authentication
• 📊 Prometheus metrics + Grafana dashboards for observability
• 🐳 Dockerized with docker-compose for local orchestration
• 📓 Dedicated notebooks/ and training/ folders for model development

Project Structure

.
├── app/
│   ├── api/            # Route definitions (auth, predict, etc.)
│   ├── cache/          # Redis cache utilities
│   ├── core/           # App-wide config, security, exceptions
│   ├── middleware/      # Custom middleware
│   ├── models/         # Serialized ML model artifacts (e.g. model.joblib)
│   ├── services/        # Business logic (model inference, etc.)
│   └── main.py          # FastAPI app entrypoint
├── data/                 # Datasets used for training
├── notebooks/            # Exploratory analysis & experimentation
├── training/             # Model training scripts/pipeline
├── Dockerfile
├── docker-compose.yml
├── prometheus.yml
├── pyproject.toml
├── requirements.txt
├── uv.lock
└── README.md

Prerequisites

• Python 3.12
• Docker & Docker Compose (recommended for local development)
• uv (optional, for dependency management via pyproject.toml / uv.lock)

Getting Started

Option 1 — Run with Docker Compose (recommended)

This spins up the API alongside Redis, Prometheus, and Grafana.

docker-compose up --build

The API will be available at http://localhost:8000.

Option 2 — Run locally

1. Clone the repository:

   git clone https://github.com/ayushagarwal27/fastapi-ml-project.git
   cd fastapi-ml-project

2. Create a virtual environment and install dependencies:

   python -m venv .venv
   source .venv/bin/activate  # on Windows: .venv\Scripts\activate
   pip install -r requirements.txt

   Or with uv:

   uv sync

3. Set up environment variables (see Configuration below).

4. Run the app:

   uvicorn app.main:app --host 0.0.0.0 --port 8000 --reload

5. Visit the interactive API docs at http://localhost:8000/docs.

Configuration

The app reads configuration from environment variables (see app/core/config.py). Create a .env file in the project root:

API_KEY=your-api-key
JWT_SECRET_KEY=your-jwt-secret
REDIS_URL=redis://localhost:6379
MODEL_PATH=app/models/model.joblib

Variable	Description	Default
API_KEY	API key for request authentication	demo-key
JWT_SECRET_KEY	Secret used to sign JWTs	jwt-secret-key
REDIS_URL	Redis connection string	redis://localhost:6379
MODEL_PATH	Path to the serialized model file	app/models/model.joblib

API Endpoints

Method	Endpoint	Description
POST	/auth/login	Authenticate and receive a JWT
POST	/predict	Predict car price from input features

Full request/response schemas are available via the auto-generated Swagger docs at /docs.

Example: Predict Car Price

curl -X POST http://localhost:8000/predict \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer <token>" \
  -d '{
    "make": "Toyota",
    "model": "Corolla",
    "year": 2019,
    "mileage": 42000,
    "fuel_type": "Petrol"
  }'

Predictions are cached in Redis based on the input payload, so repeated requests with identical inputs return instantly from cache.

Model Training

Training scripts and notebooks live in training/ and notebooks/. To retrain and export a new model:

python training/train.py

This should produce a new model.joblib file — update MODEL_PATH (or overwrite the existing file at app/models/model.joblib) to use it.

Monitoring

• Prometheus scrapes metrics from the API using the configuration in prometheus.yml.
• Grafana can be pointed at the Prometheus data source to visualize request latency, throughput, and error rates.

When running via docker-compose, Prometheus and Grafana are started alongside the API automatically.

Tech Stack

• FastAPI — web framework
• joblib — model serialization/loading
• Redis — caching layer
• Prometheus / Grafana — monitoring & observability
• Docker / Docker Compose — containerization

Contributing

Issues and pull requests are welcome. Please open an issue first to discuss any significant changes.

License

Specify your license here (e.g. MIT).