MLServe is a lightweight FastAPI app that allows you to serve and interact with a machine learning model using REST endpoints. It loads (or trains) a RandomForest model on the Iris dataset and makes predictions via a /predict API.

• Loads or trains an ML model (RandomForestClassifier)
• Accepts feature input as JSON
• Returns prediction from deployed model
• Includes Swagger UI

• FastAPI
• scikit-learn
• joblib
• Docker (optional)

pip install -r requirements.txt
uvicorn main:app --reload

docker build -t mlserve .
docker run -p 8000:8000 mlserve

POST /predict
Content-Type: application/json

{
  "features": [5.1, 3.5, 1.4, 0.2]
}

Returns:

{
  "prediction": 0
}

© 2025