deploy-inference.yaml

name: Deploy MLOps Inference Endpoint  and Monitoring Stack
on:
  push: 
   branches:
   - main

env:
  AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
  AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
  region: "ap-south-1"
  AWS_DEFAULT_REGION: "ap-south-1"

jobs:
  build:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/setup-python@v2
        with:
          python-version: 3.10.8

      - name: Checkout the repo code
        uses: actions/checkout@v3
        with:
          path: pose-estimation
          clean: true

      - name: Set AWS credentials as environment variables
        run: |
          echo "Setting AWS credentials as environment variables"
          export AWS_ACCESS_KEY_ID=$AWS_ACCESS_KEY_ID
          export AWS_SECRET_ACCESS_KEY=$AWS_SECRET_ACCESS_KEY
          export region=$region

      - name: Install dependencies
        run: |
          python -m pip install --upgrade pip
          cd pose-estimation
          pip install -r deployment/requirements.txt

      - name: Trigger Real-time Inferencing Pipeline for Pre-Trained Model deployment
        run: |
          cd pose-estimation
          python3 deployment/realtime_endpoint_deployment/realtime_endpoint_deployment.py

      - name: Attaching Model Monitoring Pipeline for Real-time Inferencing end point
        run: |
          cd pose-estimation
          python3 deployment/model_monitoring_deployment/model_monitoring_deployment.py

      - name: Install AWS CLI
        run: |
          sudo apt-get update
          sudo apt-get install -y awscli

      - name: Run CloudFormation Stack - API Gateway & Lambda
        env:
          AWS_REGION: ap-south-1
          STACK_NAME: mlops-pipeline-human-pose-prediction
        run: |
          export AWS_ACCESS_KEY_ID=$AWS_ACCESS_KEY_ID
          export AWS_SECRET_ACCESS_KEY=$AWS_SECRET_ACCESS_KEY
          export region=$region
          cd pose-estimation/aws_stack
          aws cloudformation create-stack --stack-name $STACK_NAME --region $AWS_REGION --template-body file://gateway_lamda_s3_website_creation.yml --capabilities CAPABILITY_IAM --parameters file://parameters.json

      - name: Wait for stack to complete - API Gateway & Lambda
        env:
          AWS_REGION: ap-south-1
          STACK_NAME: mlops-pipeline-human-pose-prediction
        run: |
          aws cloudformation wait stack-create-complete --stack-name $STACK_NAME --region $AWS_REGION || exit 1 # Fail if stack creation failed

      - name: Stack Output & Update HTML
        env:
          STACK_NAME: mlops-pipeline-human-pose-prediction
        run: |
          python -m pip install --upgrade pip
          cd pose-estimation
          pip install -r deployment/requirements.txt
          python3 deployment/update_html.py "$(aws cloudformation describe-stacks --stack-name $STACK_NAME --query 'Stacks[0].Outputs[0].['OutputValue'][0]' --output json)"
          
      - name: Run CloudFormation Stack - Publish Public Human Pose Prediction WebSite
        env:
          AWS_REGION: ap-south-1
          STACK_NAME: website
          BUCKET_NAME: mlops-pipeline-humanpose-estimation-prediction-demo
          AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID_GLOBAL_S3 }}
          AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY_GLOBAL_S3 }}
        run: |
          export AWS_ACCESS_KEY_ID=$AWS_ACCESS_KEY_ID
          export AWS_SECRET_ACCESS_KEY=$AWS_SECRET_ACCESS_KEY
          export region=$region

          cd pose-estimation/aws_stack
          aws cloudformation create-stack --stack-name $STACK_NAME --region $AWS_REGION --template-body file://stack_static_website_s3_public.yml --capabilities CAPABILITY_IAM --parameters ParameterKey=s3Bucketname,ParameterValue=$BUCKET_NAME
          aws cloudformation wait stack-create-complete --stack-name $STACK_NAME --region $AWS_REGION || exit 1 # Fail if stack creation failed      
          cd ../website
          aws s3 cp index.html s3://$BUCKET_NAME/index.html
          aws s3 cp error.html s3://$BUCKET_NAME/error.html
          output=$(aws cloudformation describe-stacks --stack-name $STACK_NAME --query 'Stacks[0].Outputs[1].['OutputValue'][0]' --output text)
          echo "***************HUMAN POSE PREDICTION ENDPOINT URL***************: $output"