1. The Problem
A .pkl model is dead weight without a way to interact with it. You need a bridge: an API.
2. The What (Plain English)
A web API is the front door to your model. Tools:
- FastAPI: defines endpoints.
- Uvicorn: runs the server.
Goal: prove your stack works by serving a basic “Hello, World!”
3. The How (Lab Notebook)
Step A: Install Dependencies
pip install fastapi "uvicorn[standard]"Step B: Hello World App
app.py:
from fastapi import FastAPI
app = FastAPI()
@app.get("/")
def read_root():
return {"message": "Welcome to the Clarion API"}Step C: Run the Server
uvicorn app:app --reloadVisit http://127.0.0.1:8000 — you should see your JSON response.
4. Academic Bridge
-
Cloud Unit 1.4: Cloud Service Models — simple PaaS/SaaS example.
-
OS Unit 2.1: Processes — running Uvicorn spawns a server process.
5. Why It Matters
Serving models via APIs is the backbone of MLOps. Knowing how to ship a model from notebook to production makes you valuable.