apiVersion: apps/v1
kind: Deployment
metadata:
  name: openclaw-brain-vllm
  namespace: customer1
  labels:
    app: openclaw-brain
spec:
  replicas: 1
  selector:
    matchLabels:
      app: openclaw-brain
  template:
    metadata:
      labels:
        app: openclaw-brain
    spec:
      nodeSelector:
        cloud.google.com/gke-accelerator: "nvidia-rtx-pro-6000"
      tolerations:
        - key: "nvidia.com/gpu-nvidia-rtx-pro-6000"
          operator: "Equal"
          value: "present"
          effect: "NO_SCHEDULE"
      containers:
        - name: vllm-gemma
          image: vllm/vllm-openai:latest
          command: ["python3", "-m", "vllm.entrypoints.openai.api_server"]
          args:
            - "--model"
            - "google/gemma-2b-it" # distilled gemma model 
            - "--dtype"
            - "bfloat16"
            - "--max-model-len"
            - "8192"
          ports:
            - containerPort: 8000
          resources:
            limits:
              nvidia.com/gpu: 1
              memory: "32Gi"
              cpu: "8"
            requests:
              nvidia.com/gpu: 1
              memory: "16Gi"
              cpu: "4"
          volumeMounts:
            - name: model-cache
              mountPath: /root/.cache/huggingface
      volumes:
        - name: model-cache
          emptyDir: {}
---
apiVersion: v1
kind: Service
metadata:
  name: openclaw-brain-service
  namespace: customer1
spec:
  selector:
    app: openclaw-brain
  ports:
    - protocol: TCP
      port: 8000
      targetPort: 8000
  type: ClusterIP