Chapter 8: Deploy Your AI — Get a Public URL

In this chapter

• Avani from Bengaluru ships her first AI to 1,000 users.
• Streamlit — turn any Python ML script into a web app.
• Hugging Face Spaces — free hosting your friends can visit.
• Save and load models with pickle and joblib.
• Get your own yourname.streamlit.app or HF Spaces URL.
• Put your live URL on your CV. Today.

Streamlit — Python to web app

Streamlit is a free Python library that turns ML scripts into interactive web apps. You write Python; Streamlit handles the buttons, sliders, charts and file uploads. The same code works on your laptop and on a public URL. It’s used by Google, Meta, ISRO, and every Indian AI student today — no HTML, no JavaScript, just Python.

💻 CODE LAB app.py — A Working ML Web App

import streamlit as st
import pandas as pd
import joblib

# Load your trained model (saved earlier with joblib.dump)
model = joblib.load('lpg_price_model.pkl')

st.title('LPG Price Predictor - India')
st.markdown('Predict next-month cylinder price for your state')

state = st.selectbox('Your state',
    ['Delhi', 'Maharashtra', 'Karnataka', 'Bihar', 'Tamil Nadu'])
current_price = st.slider('Current price (Rs)', 800, 1500)

if st.button('Predict'):
    pred = model.predict([[current_price, hash(state) % 100]])[0]
    st.success('Next month: Rs ' + str(round(pred)))
    st.balloons()

🚀 DEPLOY BOX Streamlit Cloud Deployment — 6 commands

# 1. Install Streamlit on your laptop
pip install streamlit pandas scikit-learn joblib

# 2. Save your trained model
python train.py        # creates lpg_price_model.pkl

# 3. Test app.py locally
streamlit run app.py   # opens at localhost:8501

# 4. Push to GitHub (create a repo first)
git init && git add . && git commit -m 'first AI app'
git push origin main

# 5. Go to streamlit.io/cloud, sign in with GitHub
# 6. Click 'New App', pick your repo, click 'Deploy'

# Done. Your URL: yourname-lpg-predictor.streamlit.app
# Free forever. Auto-redeploys on every git push.

⚙ TRY IT TODAY

Deploy ANY model from this book. Get your live URL. Add it to your bio. Show your parents. This is the single most important thing you will do in this entire series.

Hugging Face Spaces — the pro path

Streamlit Cloud is easiest for small ML apps — its free tier supports 1 GB RAM, fine for sklearn models. Hugging Face Spaces is more powerful, supports GPU (free up to 16 GB RAM), and is better for deep learning, neural networks and transformers. Every serious Indian AI shipped publicly today is on HF Spaces. Either works for your first app — start with Streamlit, and upgrade to HF when you build a neural net.

🚀 DEPLOY BOX HF Spaces Deployment

# 1. Make an account at huggingface.co
# 2. Click 'New Space', pick a name + Streamlit SDK

# 3. Clone the space to your laptop
git clone https://huggingface.co/spaces/USERNAME/SPACE
cd SPACE

# 4. Copy your app.py and model.pkl into this folder
cp ~/Desktop/app.py .
cp ~/Desktop/lpg_price_model.pkl .

# 5. Add a requirements.txt
# (list: streamlit, scikit-learn, joblib, pandas)

# 6. Push
git add . && git commit -m 'deploy' && git push

# Your URL: huggingface.co/spaces/USERNAME/SPACE

Questions — Chapter 8

1. List the 6 Streamlit deployment commands in order.
2. What is in a requirements.txt and why does HF Spaces need it?
3. What is the difference between localhost:8501 and your public URL?
4. Sketch the screen of an AI app you would build.
5. Why does shipping to a URL make you an engineer, not just a student?

Check your understanding — Chapter 8

Tap an answer to check it instantly. These are for practice — nothing is saved or sent.