Model Serving

Overview

Serving turns a saved model into a reliable, low-latency API. The concerns shift from accuracy to latency, throughput, robustness, and observability. The model artifact and its preprocessing must travel together (use the pipeline from sklearn-pipelines).

When to use

• A validated model needs to be callable by other systems.
• Moving from notebook to production.

Minimal FastAPI service

from fastapi import FastAPI
from pydantic import BaseModel
import joblib, numpy as np