AI Model Testing Project

This project demonstrates a complete workflow for developing, testing, and deploying an AI model. The primary objective is to build and test a machine learning model, expose it via a REST API, and automate the entire process using CI/CD pipelines. Below is a detailed summary of the work done.

---

Table of Contents

1. Project Description
2. Setup Instructions
3. Key Features
4. Step-by-Step Workflow
5. CI/CD Pipeline
6. API Usage
7. Technologies Used
8. License

---

Project Description

This project involves:

1. Building a machine learning model to predict survival on the Titanic dataset.
2. Automating testing for the model’s functionality and accuracy using PyTest.
3. Exposing the trained model as a REST API using Flask.
4. Containerizing the application using Docker.
5. Automating CI/CD workflows using GitHub Actions.

---

Setup Instructions

Pre-requisites

• Python 3.9 or above
• Docker installed on your machine
• A GitHub account
• Optional: Docker Hub account

Setup the Project

1. Clone the repository:

   git clone <repository-url>
   cd ai-model-testing

2. Install dependencies:

   pip install -r requirements.txt

3. Run the Flask application locally:

   python app.py

4. Test the application locally:

   pytest

5. Build and run the Docker container:

   docker build -t ai-model-api .
   docker run -p 5000:5000 ai-model-api

---

Key Features

• Machine Learning Model: Logistic regression on Titanic dataset.
• Testing: Automated testing with PyTest.
• API: Flask-based REST API for model predictions.
• Containerization: Dockerized application.
• CI/CD: GitHub Actions for automation.

---

Step-by-Step Workflow

1. Data Loading and Preprocessing

• Loaded the Titanic dataset.
• Cleaned data by handling missing values and encoding categorical variables.
• Prepared the data for training.

2. Model Training and Evaluation

• Trained a logistic regression model.
• Evaluated accuracy to ensure the model meets performance criteria.

3. Automated Testing

• Wrote PyTest scripts to:
  • Validate data preprocessing.
  • Ensure model accuracy meets thresholds.
• Integrated tests into the CI/CD pipeline.

4. Exposing the Model as an API

• Created a Flask REST API with an endpoint to make predictions.
• Tested the API locally with tools like Postman and cURL.

5. Dockerization

• Wrote a Dockerfile to containerize the application.
• Built and ran the Docker container locally.
• Pushed the container image to Docker Hub.

6. CI/CD Pipeline

• Configured GitHub Actions to:
  • Run automated tests.
  • Generate and upload test/coverage reports.
  • Build and push the Docker image to Docker Hub.
• Verified the pipeline’s success on every commit.

---

CI/CD Pipeline

The GitHub Actions pipeline automates the following steps:

1. Checkout Code:
   • Clones the repository.
2. Set Up Python:
   • Installs dependencies from requirements.txt.
3. Run Tests:
   • Executes PyTest scripts.
   • Generates test and coverage reports.
4. Build and Push Docker Image:
   • Logs into Docker Hub using GitHub secrets.
   • Builds and pushes the Docker image.

---

API Usage

Endpoint

• URL: http://<host>:5000/predict
• Method: POST

Input Example

{
  "Pclass": 3,
  "Sex": 1,
  "Age": 25,
  "SibSp": 0,
  "Parch": 0,
  "Fare": 7.25,
  "Embarked": 2
}

Response Example

{
  "prediction": 0
}

---

Technologies Used

• Programming Language: Python 3.9
• Libraries: Flask, Scikit-learn, PyTest
• Containerization: Docker
• CI/CD: GitHub Actions

---

License

This project is licensed under the MIT License. See the LICENSE file for details.

---

Next Steps

1. Explore AI model testing techniques:
   • Functional testing.
   • Performance testing.
   • Bias and fairness testing.
2. Deploy the container on a cloud platform.
3. Integrate notifications into the CI/CD pipeline.

---