apiVersion: apps/v1
kind: Deployment
metadata:
  name: openclaw-brain-vllm-l4
  namespace: customer1
  labels:
    app: openclaw-brain-l4
spec:
  replicas: 1
  selector:
    matchLabels:
      app: openclaw-brain-l4
  template:
    metadata:
      labels:
        app: openclaw-brain-l4
    spec:
      nodeSelector:
        workload: "llm-analyst"
      tolerations:
        - key: "nvidia.com/gpu"
          operator: "Equal"
          value: "present"
          effect: "NoSchedule"
      containers:
        - name: vllm-brain-l4
          image: vllm/vllm-openai:latest
          command: ["python3", "-m", "vllm.entrypoints.openai.api_server"]
          env:
            - name: HF_TOKEN
              value: ""
            - name: VLLM_LOGGING_LEVEL
              value: DEBUG
          args:
            - --model=p-e-w/Qwen3-8B-heretic
            
            - --host=0.0.0.0
            - --port=8000
            - --tensor-parallel-size=1
            - --gpu-memory-utilization=0.95
            - --max-model-len=32768
            - --kv-cache-dtype=fp8 
            - --max-num-batched-tokens=2048
            - --max-num-seqs=128
            - --enable-chunked-prefill
            - --enable-auto-tool-choice 
            - --trust-remote-code
            - --dtype=auto
            - --tool-call-parser=hermes 
            - --enable-prefix-caching
            
            
          ports:
            - containerPort: 8000
          resources:
            limits:
              nvidia.com/gpu: 1
              memory: "30Gi"
              cpu: "8"
            requests:
              nvidia.com/gpu: 1
              memory: "16Gi"
              cpu: "4"
          volumeMounts:
            - name: model-cache
              mountPath: /root/.cache/huggingface
      volumes:
        - name: model-cache
          persistentVolumeClaim:
            claimName: vllm-model-cache-pvc-l4
---
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: vllm-model-cache-pvc-l4
  namespace: customer1
spec:
  accessModes:
    - ReadWriteOnce
  resources:
    requests:
      storage: 50Gi
---
apiVersion: v1
kind: Service
metadata:
  name: openclaw-brain-service-l4
  namespace: customer1
spec:
  selector:
    app: openclaw-brain-l4
  ports:
    - protocol: TCP
      port: 8000
      targetPort: 8000
  type: ClusterIP