FastAPI for model serving in ML Python

Introduction

FastAPI helps you quickly share your machine learning model so others can use it easily through the internet.

You want to let a website or app get predictions from your model.

You need to test your model with real data from users.

You want to share your model with teammates without sending code.

You want to build a simple API to get model results anytime.

You want to deploy your model on a server for others to access.

Syntax

ML Python

from fastapi import FastAPI
from pydantic import BaseModel

app = FastAPI()

class InputData(BaseModel):
    feature1: float
    feature2: float

@app.post('/predict')
async def predict(data: InputData):
    # Use your model here
    prediction = data.feature1 + data.feature2
    return {'prediction': prediction}

Use @app.post to create a POST endpoint where you send data.

Define input data with BaseModel to check data types automatically.

Examples

A simple GET endpoint that returns a welcome message.

ML Python

from fastapi import FastAPI

app = FastAPI()

@app.get('/')
async def root():
    return {'message': 'Hello World'}

Example of a POST endpoint that calculates a score from input features.

ML Python

from fastapi import FastAPI
from pydantic import BaseModel

app = FastAPI()

class InputData(BaseModel):
    age: int
    income: float

@app.post('/predict')
async def predict(data: InputData):
    score = data.age * 0.1 + data.income * 0.01
    return {'score': score}

Sample Model

This program creates a FastAPI app with one POST endpoint '/predict'. It takes two numbers, adds them, and returns the result as prediction.

Run this code, then send a POST request with JSON like {"feature1": 3.5, "feature2": 2.5} to get prediction 6.0.

ML Python

from fastapi import FastAPI
from pydantic import BaseModel
import uvicorn

app = FastAPI()

class InputData(BaseModel):
    feature1: float
    feature2: float

@app.post('/predict')
async def predict(data: InputData):
    # Simple model: sum of features
    prediction = data.feature1 + data.feature2
    return {'prediction': prediction}

if __name__ == '__main__':
    uvicorn.run(app, host='127.0.0.1', port=8000)

Important Notes

Use tools like curl or Postman to send POST requests to your FastAPI server.

FastAPI automatically creates interactive docs at /docs to test your API.

Keep your model loading outside the endpoint function to avoid reloading on every request.

Summary

FastAPI makes it easy to share your ML model as a web service.

Define input data clearly with Pydantic models for safety and clarity.

Use POST endpoints to send data and get predictions back in JSON format.