REST API serving with FastAPI in MLOps - Mini Project: Build & Apply

REST API serving with FastAPI

📖 Scenario: You are building a simple REST API to serve predictions from a machine learning model. This API will receive input data and return a prediction result. FastAPI is a modern, fast web framework for Python that makes it easy to create APIs.

🎯 Goal: Build a basic REST API using FastAPI that accepts input data via a POST request and returns a prediction result as JSON.

📋 What You'll Learn

Create a FastAPI app instance

Define a Pydantic model for input data validation

Create a POST endpoint '/predict' that accepts input data

Return a JSON response with a prediction message

💡 Why This Matters

🌍 Real World

Serving machine learning model predictions via REST APIs is common in production systems. FastAPI helps build these APIs quickly and efficiently.

💼 Career

Understanding how to serve ML models with APIs is essential for MLOps engineers and data scientists working on deploying models to production.

1

Create FastAPI app and input data model

Import FastAPI and BaseModel from fastapi and pydantic respectively. Create a FastAPI app instance called app. Define a Pydantic model called InputData with one field feature of type float.

MLOps

# Your code here

Need a hint?

Use app = FastAPI() to create the app. Define InputData class inheriting from BaseModel with a feature attribute.

2

Add a POST endpoint for prediction

Add a POST endpoint /predict to the app using the @app.post("/predict") decorator. Define a function predict that takes one parameter input_data of type InputData.

MLOps

from fastapi import FastAPI
from pydantic import BaseModel

app = FastAPI()

class InputData(BaseModel):
    feature: float

# Add POST endpoint /predict below
# Your code here

Need a hint?

Use @app.post("/predict") decorator and define predict function with input_data: InputData parameter.

3

Implement prediction logic

Inside the predict function, create a variable result that multiplies input_data.feature by 2. Return a dictionary with key prediction and value as the result.

MLOps

from fastapi import FastAPI
from pydantic import BaseModel

app = FastAPI()

class InputData(BaseModel):
    feature: float

@app.post("/predict")
def predict(input_data: InputData):
    # Your code here

Need a hint?

Calculate result by multiplying input_data.feature by 2. Return it in a dictionary with key prediction.

4

Run the FastAPI app and test output

Print the string "FastAPI app ready to serve predictions" to confirm the app setup is complete.

MLOps

from fastapi import FastAPI
from pydantic import BaseModel

app = FastAPI()

class InputData(BaseModel):
    feature: float

@app.post("/predict")
def predict(input_data: InputData):
    result = input_data.feature * 2
    return {"prediction": result}

# Print confirmation message
# Your code here

Need a hint?

Use print("FastAPI app ready to serve predictions") to show the message.