Explore

Building & Deploying AI Applications with CI/CD

Lab Instructions: Building & Deploying AI Applications with CI/CD

AML3304 - Software Tools for AI/ML Version 2.1 | March 2025

Lab Objectives

Implement end-to-end ML pipeline using Google Colab & Hugging Face

Configure GitHub Actions CI/CD for model updates

Manage projects via GitHub Issues + Project Boards

Deploy production-grade models to Hugging Face Spaces

Technical Stack

Table 1

Table 1

Component

Tools

Development Environment

Google Colab (Free Tier)

Model Framework

Hugging Face Transformers

CI/CD Pipeline

GitHub Actions

Deployment

Hugging Face Spaces

Project Management

GitHub Issues + Project Boards

There are no rows in this table

⁠

Step-by-Step Implementation

Phase 1: Environment Setup (1 Hour)

1.1 Create Lab Repository

bash

git clone https://github.com/[YOUR_USERNAME]/aml3304-lab.git

cd aml3304-lab

1.2 Colab Starter Notebook Create AML3304_Lab.ipynb with:

python

!pip install -q transformers==4.30 datasets==2.12 scikit-learn==1.2 mlflow==2.5

import mlflow

mlflow.set_tracking_uri("sqlite:///mlruns.db")

Phase 2: Model Development (3 Hours)

2.1 Text Classification Pipeline

python

from transformers import pipeline

# Simple sentiment analysis

classifier = pipeline("text-classification", model="distilbert-base-uncased-finetuned-sst-2-english")

# Test prediction

sample_text = "Google Colab provides excellent GPU resources for AI education"

print(classifier(sample_text))

2.2 MLflow Tracking

python

with mlflow.start_run():

mlflow.log_param("model", "distilbert-base-uncased")

mlflow.log_metric("accuracy", 0.92)

mlflow.transformers.log_model(

transformers_model=classifier,

artifact_path="sentiment_model",

registered_model_name="lab-sentiment-v1"

)

Phase 3: Deployment (2 Hours)

3.1 Hugging Face Spaces Setup

Create new Space at

huggingface.co/spaces⁠

⁠

Select Gradio SDK

3.2 Deployment Script Create app.py:

python

import gradio as gr

from transformers import pipeline

model = pipeline("text-classification", "distilbert-base-uncased-finetuned-sst-2-english")

def analyze(text):

return model(text)[0]

gr.Interface(

fn=analyze,

inputs=gr.Textbox(lines=2, placeholder="Enter text..."),

outputs="label"

).launch()

Phase 4: CI/CD Implementation (2 Hours)

4.1 GitHub Actions Workflow Create .github/workflows/deploy.yml:

text

name: Deploy to Hugging Face

on:

push:

branches: [ main ]

jobs:

build-deploy:

runs-on: ubuntu-latest

steps:

- uses: actions/checkout@v3

- name: Set up Python

uses: actions/setup-python@v4

with:

python-version: '3.10'

- name: Install dependencies

run: |

pip install transformers gradio

- name: Deploy to Spaces

uses: huggingface/huggingface_hub@main

with:

endpoint: spaces

hf_token: ${{ secrets.HF_TOKEN }}

git_user_name: "AML3304 Student"

git_user_email: "student@lambtoncollege.ca"

4.2 Project Management Setup

Enable GitHub Projects in repo settings

Create template issues:

/kind bug

/kind enhancement

/kind documentation

Testing & Validation

Sample Test Suite Create tests/test_pipeline.py:

python

def test_data_loading():

from datasets import load_dataset

data = load_dataset("imdb", split="test[:1%]")

assert len(data) > 0

def test_model_inference():

from transformers import pipeline

classifier = pipeline("text-classification", model="distilbert-base-uncased-finetuned-sst-2-english")

result = classifier("This lab works perfectly!")

assert result[0]['label'] in ['POSITIVE', 'NEGATIVE']

Monitoring & Maintenance

GitHub Actions Dashboard: Monitor build statuses

MLflow UI:

bash

mlflow ui --backend-store-uri sqlite:///mlruns.db

Hugging Face Metrics: Track API usage & performance

Ethical Implementation Checklist

Bias evaluation using datasets library

Privacy filter for user inputs

Model card documentation

Submission Requirements

Functional Colab notebook

GitHub repo with:

Passing CI/CD workflows

Active project board with ≥5 issues

Live Hugging Face Space demo

Recommended Resources

⁠

Hugging Face Course⁠

⁠

GitHub Actions Docs⁠

⁠

MLflow Tracking Guide⁠

⁠

This lab aligns with Vocational Learning Outcomes 3, 6, and 9 by implementing end-to-end MLOps workflows with industry-standard tools.

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.