---

CI/CD Deployment of Machine Learning Model to GCP (MLOPS)

This project provides an automated CI/CD pipeline to deploy machine learning models to Google Cloud Platform (GCP) using Google Cloud Run and Artifact Registry. The pipeline leverages GitHub Actions to automate the entire process, from building a Docker container to deploying the model on GCP.

---

Key Features

• Automated Deployment: CI/CD pipeline that triggers on every push to the main branch or pull request.
• Dockerized ML Models: The model is containerized with Docker to ensure consistent deployment across environments.
• Cloud Run Integration: Deploys the Docker image to Google Cloud Run, enabling scalable, serverless model serving.
• Versioned Images: Each deployment is tagged with the commit hash to ensure version control of Docker images.
• Efficient Dependency Management: The workflow optimizes the installation and caching of dependencies to speed up builds.

---

Architecture Overview

1. GitHub Actions Workflow: Automates the CI/CD process on every push or pull request.
2. Docker Image Creation: The machine learning model and dependencies are containerized using Docker.
3. Artifact Registry: The Docker image is pushed to Google Cloud’s Artifact Registry for secure storage.
4. Cloud Run Deployment: The image is deployed to Google Cloud Run for seamless, scalable, serverless serving of the model.

---

Workflow Steps

1. Code Checkout

• GitHub Actions checks out the latest code from the repository.

2. Install Dependencies

• Installs both Python and system-level dependencies required for the machine learning model and training environment.

3. Run Data Pipeline

• Data Ingestion: Collects raw data necessary for training.
• Data Processing: Cleans, transforms, and prepares the data.
• Model Training: Trains the machine learning model.

4. Docker Image Build and Push

• Builds the Docker image and tags it with the Git commit hash for versioning.
• Pushes the Docker image to Google Cloud Artifact Registry.

5. Deploy to Cloud Run

• Deploys the Docker image to Google Cloud Run, making the model available as a fully managed, scalable web service.

---

Prerequisites

Before using this pipeline, ensure that you have the following:

1. Google Cloud Project: Set up a GCP project where Cloud Run and Artifact Registry will be used.
2. Google Cloud SDK: Required for managing GCP services like Cloud Run and Artifact Registry.
3. Service Account Key: Create a service account in GCP with the required permissions (roles/cloudrun.admin, roles/artifactregistry.writer, etc.) and store the key as a GitHub secret.

---

Setup

1. Clone the Repository

git clone https://github.com/your-username/ml-cicd-deployment-gcp.git
cd ml-cicd-deployment-gcp

2. Add Google Cloud Service Account Key to GitHub Secrets

• In your GitHub repository, go to Settings > Secrets.

• Add the following secrets:

  • GCP_PROJECT_ID: Your GCP project ID.
  • GCP_SA_KEY: The service account key (in JSON format) for GCP authentication.

3. Customize Your GCP Settings

• Update the REGION and ARTIFACT_REGISTRY_URL in the workflow file (.github/workflows/ml_pipeline.yml) to match your GCP setup.

REGION: us-central1  # or your preferred GCP region
ARTIFACT_REGISTRY_URL: us-central1-docker.pkg.dev/${{ secrets.GCP_PROJECT_ID }}/my-repo/credit-card-risk-prediction

4. Commit and Push Changes

git add .
git commit -m "Set up CI/CD pipeline for ML model deployment"
git push origin main

5. Trigger the CI/CD Pipeline

The GitHub Actions workflow will automatically trigger on the main branch, starting from code checkout to deployment on Google Cloud Run.

---

Deployment Details

• Cloud Service: The model is deployed to Google Cloud Run, which automatically scales your service based on incoming traffic. It handles the heavy lifting of provisioning and managing infrastructure for you.
• Authentication: The service uses the service account key stored in GitHub Secrets for secure authentication to Google Cloud services.

---

Troubleshooting

• Logs: You can check the logs of your Cloud Run deployment by visiting the Google Cloud Console under Cloud Run > [Your Service] > Logs.
• Timeouts: If the container fails to start within the allocated timeout, try increasing the timeout duration or check the port configuration (default is PORT=8080).
• Permission Issues: Ensure that your service account has the required permissions to access Cloud Run, Artifact Registry, and other necessary resources.

For more details on troubleshooting Cloud Run deployments, visit the GCP troubleshooting documentation.

---

Contributing

Feel free to fork this repository, submit issues, or create pull requests. Contributions are welcome to enhance the functionality, optimize the workflow, or fix bugs!

---

License

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

---

Acknowledgments

• GitHub Actions: For continuous integration and deployment.
• Docker: For containerizing the machine learning models.
• Google Cloud Platform (GCP): For providing scalable infrastructure with Cloud Run and Artifact Registry.
• Python Libraries: For building and training the machine learning models.

---