AI Model Deployment Hands-on Exercise + Best Practices Recap

Hands-on Exercise: Deploying a Simple AI Model

In this exercise, you will deploy a basic AI model and interact with it through an API. The goal is to get hands-on experience with deploying models in a simple and reproducible way using tools available in your Docker-based environment.

Step 1: Train and Save a Model Using Scikit-learn

We’ll train a simple machine learning model using scikit-learn and save it for deployment.

import pandas as pd
import numpy as np
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import train_test_split
from sklearn.datasets import load_iris
from sklearn.metrics import accuracy_score
import joblib
import mlflow
import mlflow.sklearn

data = load_iris()
df = pd.DataFrame(data.data, columns=data.feature_names)
labels = data.target

X_train, X_test, y_train, y_test = train_test_split(df, labels, test_size=0.2, random_state=42)

with mlflow.start_run() as run:
    n_estimators = 100
    random_state = 42

    mlflow.log_param("n_estimators", n_estimators)
    mlflow.log_param("random_state", random_state)
    print(f"Logged parameters: n_estimators={n_estimators}, random_state={random_state}")

    model = RandomForestClassifier(n_estimators=n_estimators, random_state=random_state)
    model.fit(X_train, y_train)

    y_pred = model.predict(X_test)
    accuracy = accuracy_score(y_test, y_pred)

    mlflow.log_metric("accuracy", accuracy)
    print(f"Logged metric: accuracy={accuracy:.4f}")

    mlflow.sklearn.log_model(
        sk_model=model,
        artifact_path="iris-random-forest-model",
    )
    print("Model logged to MLflow Tracking Server under run_id:", run.info.run_id)

    local_save_path = "sklearn_iris_model_local"
    mlflow.sklearn.save_model(model, local_save_path)
    print(f"Model also saved locally at: {local_save_path}")

print("MLflow run completed.")

Step 2: Deploy the Model Using MLflow

Open a new terminal and run this command.

sudo docker exec -it -w /app sa-course-labs mlflow models serve -m ./sklearn_iris_model_local --port 5001 --no-conda

This will launch a REST API that can be used to make predictions.

Step 3: Make a Prediction Request

You can send a request to the deployed model using Python:

import requests
import json
from sklearn.datasets import load_iris

data_loader = load_iris()

url = "http://127.0.0.1:5001/invocations"

data_payload = {
    "dataframe_split": {
        "columns": data_loader.feature_names,
        "data": [[5.1, 3.5, 1.4, 0.2]]
    }
}

try:
    response = requests.post(url, json=data_payload, headers={"Content-Type": "application/json"}, timeout=10)
    response.raise_for_status()
    print("Prediction Response:", response.json())
except requests.exceptions.RequestException as e:
    print(f"Error making prediction request: {e}")
except Exception as e:
    print(f"An unexpected error occurred: {e}")

**Step 4: Stop the terminal opened on step 2 with Ctrl-C

Best Practices Recap

Key Takeaways for AI Model Deployment