1. Project Introduction

Upload an image and discover celebrities with AI-powered recognition and interactive Q&A.

Demo:

CI/CD Pipeline:

GKE Deployment Dashboard:

• Tech stack

  • Groq --> LLM
  • Gitlab --> Serve as SCM for the project. Also a CI/CD tool for making pipelines
  • OpenCV Python --> To deal with image work like conversion of images, scaling, etc
  • Flask --> Application backend
  • HTML/CSS --> To make UI or frontend of the app
  • Docker --> For containerization of the app during deployment
  • GCP GAR --> For storing the Docker images
  • GCP GKE --> To deploy and run the app on the cloud within a Kubernetes Cluster. It's a service offered by Google Cloud

• Workflow

2. Configure VM Instance

✅ Pre-Deployment Checklist

Check that you have done the following before moving to the CircleCI deployment part:

1. ✅ Dockerfile
2. ✅ Kubernetes Deployment file
3. ✅ Code Versioning using GitLab

🚀 GitLab Project Setup and Code Versioning

Make an account on GitLab, then create a public project and follow all necessary steps mentioned in the course video.

Now, perform code versioning using GitLab by running the following commands:

git init
git branch -M main
git remote add origin https://gitlab.com/your-username/your-repo.git
git add .
git commit -m "Initial commit"
git push origin main

✅ Enable Required GCP APIs

Go to your GCP account and enable all the following APIs:

Navigation:
In the left pane → APIs & Services → Library

Enable the following:

• Kubernetes Engine API
• Container Registry API
• Compute Engine API
• Cloud Build API
• Cloud Storage API
• IAM API

✅ Create GKE Cluster and Artifact Registry

1. Create GKE Cluster:

   • Go to your GCP Console and search for GKE.
   • Create a new cluster with a name of your choice.
   • In the Networking tab, provide the necessary access/configurations.

2. Create Artifact Registry:

   • In the GCP Console, search for Artifact Registry.
   • Create a new Artifact Registry with a name of your choice.

✅ Create a Service Account and Configure Access

1. Create a Service Account in your GCP Console.

2. Assign the following roles:

   • Storage Object Admin
   • Storage Object Viewer
   • Owner
   • Artifact Registry Admin
   • Artifact Registry Writer

3. Download the key as a .json file.

4. Place the key file (e.g., gcp-key.json) in the root directory of your project.

5. Important: Add gcp-key.json to your .gitignore to prevent it from being pushed to GitHub.

🔐 Add gcp-key.json as a GitLab CI/CD Secret Variable

1. Go to your GitLab project → Settings → CI/CD → expand the Variables section.

2. Click Add Variable with the following details:

• Key: GCP_SA_KEY

• Value:
  Run the following command in your terminal to get the base64 value of your gcp-key.json file:

  cat gcp-key.json | base64 -w 0

✅ Set Up LLMOps Secrets in GKE using kubectl

1. Access your GKE cluster:

   • Open your GKE Console.
   • Open the kubectl terminal.

2. Configure your local terminal to connect with your cluster:

   Run the following command, adjusting for your project details:

   gcloud container clusters get-credentials llmops-cluster1 \
   --region us-central1 \
   --project <your project name>

3. Create a Kubernetes secret to store your LLM API key:

   kubectl create secret generic llmops-secrets \
   --from-literal=GROQ_API_KEY="your_actual_groq_api_key"

   This secret will be referenced in your Kubernetes deployment file to securely fetch the GROQ_API_KEY.

⚙️ Set Up GitLab CI/CD with .gitlab-ci.yml

Create a .gitlab-ci.yml file in the root of your project directory.

✅ GitLab CI/CD setup complete.

Now Push code to Gitlab ---> Automatically triggers the pipeline or u can manually trigger it using

• Go to Left pane of Gitlab
• Go to Build --> Pipelines
• Trigger pipeline from there

• Note: Your first pipeline might fail if it is your first time in GitLab because u need to verify your account using phone number or credit card Do verify and again trigger the pipeline...

• Now go see ur app on GKE Workloads...