cloudbuild-endpoint.yaml

steps:
  - name: "gcr.io/google.com/cloudsdktool/cloud-sdk"
    entrypoint: "bash"
    id: upload-model
    timeout: 600s
    args:
      - "-c"
      - |
        gcloud ai models upload \
          --region="us-central1" \
          --container-ports=8080 \
          --container-image-uri="us-central1-docker.pkg.dev/${PROJECT_ID}/interview-ai-detector/model-prediction:latest" \
          --container-predict-route="/predict" \
          --container-health-route="/health" \
          --display-name="interview-ai-detector-model"

  - name: "gcr.io/google.com/cloudsdktool/cloud-sdk"
    entrypoint: "bash"
    id: fetch-model
    waitFor: upload-model
    timeout: 600s
    args:
      - "-c"
      - |
        MODEL_ID=$(gcloud ai models list --region=us-central1 --format="value(name)" | head -n 1)
        echo MODEL_ID > /workspace/model_id.txt

  - name: "gcr.io/google.com/cloudsdktool/cloud-sdk"
    entrypoint: "bash"
    id: create-endpoint
    waitFor: fetch-model
    args:
      - "-c"
      - |
        ENDPOINT_ID=$(gcloud ai endpoints create \
          --region="us-central1" \
          --display-name="interview-ai-detector-endpoint" \
          --format="value(name)")
        echo ENDPOINT_ID > /workspace/endpoint_id.txt

  - name: "gcr.io/google.com/cloudsdktool/cloud-sdk"
    entrypoint: "bash"
    waitFor: create-endpoint
    args:
      - "-c"
      - |
        gcloud ai endpoints deploy-model "${$(cat /workspace/endpoint_id.txt)}" \
          --region="us-central1" \
          --model="${$(cat /workspace/model_id.txt)}" \
          --display-name="interview-ai-detector-deployment" \
          --machine-type="n1-standard-4" \
          --accelerator="count=1,type=nvidia-tesla-t4" \
          --service-account="vertex-ai-user-managed-sa@steady-climate-416810.iam.gserviceaccount.com"