This project implements an automated workflow using GitHub Actions and Terraform to train a Machine Learning model, package it, and deploy it on AWS. The deployed infrastructure includes S3 for prediction storage, Lambda to serve the model, API Gateway to handle requests, and Route 53 to provide a custom URL.

. ├── README.md ├── iac │   ├── main.tf │   ├── modules │   │   ├── apigateway │   │   │   ├── main.tf │   │   │   └── outputs.tf │   │   ├── iam │   │   │   └── main.tf │   │   ├── lambda │   │   │   └── main.tf │   │   ├── route53 │   │   │   ├── main.tf │   │   │   ├── outputs.tf │   │   │   └── variables.tf │   │   └── s3 │   │   └── main.tf │   ├── terraform.tfvars │   └── variables.tf ├── ml_deployment │   ├── iris_model.joblib │   └── lambda_function.py └── ml_prediction └── iris_model.py

The GitHub Actions workflow is located in .github/workflows/deploy.yml and performs the following steps:

1. Checkout Repository: Retrieves the code from the repository.
2. Set up Python: Configures the Python 3.8 environment.
3. Install Dependencies: Installs the required dependencies (joblib, scikit-learn, boto3).
4. Train and Save Model: Executes the script to train the model and saves the iris_model.joblib file.
5. Package Lambda Function: Creates a ZIP file containing the Lambda function code and the trained model.
6. Install Terraform: Sets up Terraform.
7. Terraform Init and Apply: Deploys the infrastructure on AWS.

The workflow runs manually in the main branch. Ensure you have configured the AWS secrets (AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY) in the GitHub repository.

1. Successful Deployment: Once the deployment is successful, the API URL is shown in the output of the GitHub Actions workflow.
2. Make POST Requests: Use curl or a similar tool to send POST requests to the API URL.

curl -X POST https://api-url/predict \
    -H "Content-Type: application/json" \
    -d '{"features": [5.1, 3.5, 1.4, 0.2], "species_names": ["setosa", "versicolor", "virginica"]}'

### Expected Response

The API will respond with a URL pointing to the JSON file in the S3 bucket containing the prediction:

`{
  "url": "https://s3bucket.s3.amazonaws.com/predictions/request_id.json"
}`

Training the Model Locally
--------------------------

### Steps to Train the Model Locally

1.  **Install Dependencies**: Ensure you have `joblib` and `scikit-learn` installed. You can install them with pip:

    `pip install joblib scikit-learn`

2.  **Run the Training Script**: Navigate to the project directory and execute the training script:

    `python ml_prediction/iris_model.py`

3.  **Saved Model**: The trained model will be saved as `iris_model.joblib` in the `ml_deployment` directory.

Detailed File and Directory Structure
-------------------------------------

-   **`iac/`**: Contains the Terraform configuration for deploying the infrastructure.
    -   **`modules/`**: Reusable Terraform modules.
        -   **`apigateway/`**: API Gateway configuration.
        -   **`iam/`**: IAM configuration for Lambda permissions.
        -   **`lambda/`**: Lambda function configuration.
        -   **`route53/`**: Route 53 configuration for the custom domain.
        -   **`s3/`**: S3 bucket configuration.
-   **`ml_deployment/`**: Contains the trained model and Lambda function code.
    -   **`lambda_function.py`**: Lambda function code.
    -   **`iris_model.joblib`**: Trained model file.
-   **`ml_prediction/`**: Contains the script to train the model.
    -   **`iris_model.py`**: Model training script.