0Pricing
MLOps Academy · Lesson

Serve the Production Model

Load the staged model into your FastAPI service.

Serve the Production Model is a free MLOps Academy lesson on CoddyKit — lesson 3 of 4. You can read the complete lesson below for free — then practise it hands-on in the browser with a built-in code editor and a 24/7 AI tutor. It is part of the MLOps Academy learning path, one of 4 lessons in the course, and your progress syncs across the web and the CoddyKit app.

From Registry to API

The promoted model is sitting in the registry. Now you load it into a FastAPI service so real requests can get predictions. 🌐

Reference It by Alias

Build a model URI that points at the champion alias. Your code never hardcodes a version, so promotions just work.

MODEL_URI = "models:/churn-classifier@champion"

Load the Model

Use mlflow.pyfunc.load_model to pull the production model down into memory as a callable Python object.

import mlflow.pyfunc

model = mlflow.pyfunc.load_model(MODEL_URI)

Load Once at Startup

Loading is slow, so do it once when the app boots, not per request. Use a FastAPI lifespan handler to load before traffic arrives.

from contextlib import asynccontextmanager

@asynccontextmanager
async def lifespan(app):
    app.state.model = mlflow.pyfunc.load_model(MODEL_URI)
    yield

Create the App

Wire the lifespan into your FastAPI instance. From here on, the loaded model lives on app.state and is shared by every request.

from fastapi import FastAPI

app = FastAPI(lifespan=lifespan)

Validate the Input

Define a Pydantic model for the request body so malformed input is rejected before it ever reaches the model.

from pydantic import BaseModel

class Features(BaseModel):
    tenure: int
    monthly_charges: float

Write the Predict Endpoint

Turn the validated input into a row and call the model. The predict method returns the prediction you send back as JSON.

@app.post("/predict")
def predict(f: Features):
    pred = app.state.model.predict([[f.tenure, f.monthly_charges]])
    return {"prediction": int(pred[0])}

Add a Health Check

A tiny /health route lets orchestrators know the service is up and the model finished loading before they route traffic.

@app.get("/health")
def health():
    return {"status": "ok"}

Run the Server

Launch with uvicorn and your model is live on HTTP. Any client that can POST JSON can now get predictions.

uvicorn serve:app --host 0.0.0.0 --port 8000

Quick Check

Why load the model in a lifespan handler instead of inside the predict function?

Updating the Live Model

To ship a new model, re-point the champion alias and restart the service. It loads the new version at startup automatically.

Same Model Everywhere

Because the service pulls from the registry, dev, staging, and prod can all load the exact same artifact by alias. No copy-paste of weights.

Recap: Serving the Production Model

You loaded the champion at startup, validated input, exposed /predict and /health, and ran it with uvicorn. Your model is now a live service. ✅

Frequently asked questions

Is the “Serve the Production Model” lesson free?

Yes — the full text of “Serve the Production Model” is free to read here on the web, and the MLOps Academy course includes 4 lessons in total. To practise it interactively (a built-in code editor and a 24/7 AI tutor) and unlock the rest of the MLOps Academy course, upgrade to CoddyKit PRO.

What will I learn in “Serve the Production Model”?

Load the staged model into your FastAPI service. You practise MLOps Academy with hands-on code you run directly in the browser, and a 24/7 AI tutor answers your questions as you work through the lesson.

Do I need any experience to start MLOps Academy?

No prior experience is required. MLOps Academy on CoddyKit is structured for beginners through advanced learners; this is — lesson 3 of 4, so you can start here or from the beginning and move at your own pace.

How long does the “Serve the Production Model” lesson take?

Most CoddyKit lessons take about 5–10 minutes. Each one is bite-sized and interactive, so you make steady progress and pick up exactly where you left off across the web and the app.

Can I write and run code in this MLOps Academy lesson?

Yes. Every MLOps Academy lesson includes a built-in code editor, so you write and run real code right in your browser and get instant AI feedback — no local setup required.

All lessons in this course

  1. Train and Log to the Registry
  2. Promote the Best Model to Production
  3. Serve the Production Model
  4. Trace a Prediction Round-Trip
← Back to MLOps Academy