tutorials-and-examples/nvidia-nim/blueprints/digitalhuman/digital-human-nimbp.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: dighum-llama3-8b
spec:
  replicas: 1
  selector:
    matchLabels:
      app: dighum-llama3-8b
  template:
    metadata:
      labels:
        app: dighum-llama3-8b
    spec:
      containers:
      - name: dighum-llama3-8b
        image: nvcr.io/nim/meta/llama3-8b-instruct:1.0.3
        ports:
        - containerPort: 8000
        resources:
          requests:
            nvidia.com/gpu: 1
          limits:
            nvidia.com/gpu: 1
        volumeMounts:
          - name: dshm
            mountPath: /dev/shm
        env:
        - name: NGC_API_KEY
          valueFrom:
            secretKeyRef:
              name: ngc-api-key
              key: NGC_API_KEY
        - name: NIM_PEFT_REFRESH_INTERVAL
          value: "3600"
        livenessProbe:
          httpGet:
            path: /v1/health/ready
            port: 8000
          initialDelaySeconds: 10
          timeoutSeconds: 20
          periodSeconds: 10
          failureThreshold: 100
      imagePullSecrets:
      - name: secret-nvcr
      nodeSelector:
        cloud.google.com/gke-gpu: "true"
      volumes:
        - name: dshm
          emptyDir:
            medium: Memory
            sizeLimit: 20Gi

---
apiVersion: v1
kind: Service
metadata:
  name: dighum-llama3-8b-lb
spec:
  type: LoadBalancer
  selector:
    app: dighum-llama3-8b
  ports:
  - protocol: TCP
    port: 80
    targetPort: 8000 # The container still listens on 8000

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: dighum-embedqa-e5v5
spec:
  replicas: 1
  selector:
    matchLabels:
      app: dighum-embedqa-e5v5
  template:
    metadata:
      labels:
        app: dighum-embedqa-e5v5
    spec:
      containers:
      - name: dighum-embedqa-e5v5
        image: nvcr.io/nim/nvidia/nv-embedqa-e5-v5:1.0.1
        ports:
        - containerPort: 8000
        resources:
          requests:
            nvidia.com/gpu: 1
          limits:
            nvidia.com/gpu: 1
        volumeMounts:
          - name: dshm
            mountPath: /dev/shm
        env:
        - name: NGC_API_KEY
          valueFrom:
            secretKeyRef:
              name: ngc-api-key
              key: NGC_API_KEY
        livenessProbe:
          httpGet:
            path: /v1/health/ready
            port: 8000
          initialDelaySeconds: 10
          timeoutSeconds: 20
          periodSeconds: 10
          failureThreshold: 100
      imagePullSecrets:
      - name: secret-nvcr
      nodeSelector:
        cloud.google.com/gke-gpu: "true"
      volumes:
        - name: dshm
          emptyDir:
            medium: Memory
            sizeLimit: 16Gi

---
apiVersion: v1
kind: Service
metadata:
  name: dighum-embedqa-e5v5-lb
spec:
  type: LoadBalancer
  selector:
    app: dighum-embedqa-e5v5
  ports:
  - protocol: TCP
    port: 80
    targetPort: 8000

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: dighum-rerankqa-mistral4bv3
spec:
  replicas: 1
  selector:
    matchLabels:
      app: dighum-rerankqa-mistral4bv3
  template:
    metadata:
      labels:
        app: dighum-rerankqa-mistral4bv3
    spec:
      containers:
      - name: dighum-rerankqa-mistral4bv3
        image: nvcr.io/nim/nvidia/nv-rerankqa-mistral-4b-v3:1.0.1
        ports:
        - containerPort: 8000
        resources:
          requests:
            nvidia.com/gpu: 1
          limits:
            nvidia.com/gpu: 1
        volumeMounts:
          - name: dshm
            mountPath: /dev/shm
        env:
        - name: NGC_API_KEY
          valueFrom:
            secretKeyRef:
              name: ngc-api-key
              key: NGC_API_KEY
        livenessProbe:
          httpGet:
            path: /v1/health/ready
            port: 8000
          initialDelaySeconds: 10
          timeoutSeconds: 20
          periodSeconds: 10
          failureThreshold: 100
      imagePullSecrets:
      - name: secret-nvcr
      nodeSelector:
        cloud.google.com/gke-gpu: "true"
      volumes:
        - name: dshm
          emptyDir:
            medium: Memory
            sizeLimit: 20Gi

---
apiVersion: v1
kind: Service
metadata:
  name: dighum-rerankqa-mistral4bv3-lb
spec:
  type: LoadBalancer
  selector:
    app: dighum-rerankqa-mistral4bv3
  ports:
  - protocol: TCP
    port: 80
    targetPort: 8000

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: dighum-parakeet-asr-1-1b
spec:
  replicas: 1
  selector:
    matchLabels:
      app: dighum-parakeet-asr-1-1b
  template:
    metadata:
      labels:
        app: dighum-parakeet-asr-1-1b
    spec:
      containers:
      - name: dighum-parakeet-asr-1-1b
        image: nvcr.io/nim/nvidia/riva-asr:1.3.0
        ports:
        - containerPort: 9000
          name: http
        - containerPort: 50051
          name: grpc
        resources:
          requests:
            nvidia.com/gpu: 1
          limits:
            nvidia.com/gpu: 1
        volumeMounts:
          - name: dshm
            mountPath: /dev/shm
        env:
        - name: NGC_API_KEY
          valueFrom:
            secretKeyRef:
              name: ngc-api-key
              key: NGC_API_KEY
        - name: NIM_HTTP_API_PORT
          value: "9000"
        - name: NIM_GRPC_API_PORT
          value: "50051"
        - name: NIM_TAGS_SELECTOR
          value: "name=parakeet-1-1b-ctc-riva-en-us,mode=str"
        livenessProbe:
          httpGet:
            path: /v1/health/ready
            port: 9000
          initialDelaySeconds: 10
          timeoutSeconds: 20
          periodSeconds: 10
          failureThreshold: 100
      imagePullSecrets:
      - name: secret-nvcr
      nodeSelector:
        cloud.google.com/gke-gpu: "true"
      volumes:
        - name: dshm
          emptyDir:
            medium: Memory
            sizeLimit: 8Gi

---
apiVersion: v1
kind: Service
metadata:
  name: dighum-parakeet-asr-1-1b-lb
spec:
  type: LoadBalancer
  selector:
    app: dighum-parakeet-asr-1-1b
  ports:
  - protocol: TCP
    name: http
    port: 80
    targetPort: 9000
  - protocol: TCP
    name: grpc
    port: 50051
    targetPort: 50051

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: dighum-fastpitch-tts
spec:
  replicas: 1
  selector:
    matchLabels:
      app: dighum-fastpitch-tts
  template:
    metadata:
      labels:
        app: dighum-fastpitch-tts
    spec:
      containers:
      - name: dighum-fastpitch-tts
        image: nvcr.io/nim/nvidia/riva-tts:1.3.0
        ports:
        - containerPort: 9000
          name: http
        - containerPort: 50051
          name: grpc
        resources:
          requests:
            nvidia.com/gpu: 1
          limits:
            nvidia.com/gpu: 1
        volumeMounts:
          - name: dshm
            mountPath: /dev/shm
        env:
        - name: NGC_API_KEY
          valueFrom:
            secretKeyRef:
              name: ngc-api-key
              key: NGC_API_KEY
        - name: NIM_HTTP_API_PORT
          value: "9000"
        - name: NIM_GRPC_API_PORT
          value: "50051"
        - name: NIM_TAGS_SELECTOR
          value: "name=fastpitch-hifigan-en-us"
        livenessProbe:
          httpGet:
            path: /v1/health/ready
            port: 9000
          initialDelaySeconds: 10
          timeoutSeconds: 20
          periodSeconds: 10
          failureThreshold: 100
      imagePullSecrets:
      - name: secret-nvcr
      nodeSelector:
        cloud.google.com/gke-gpu: "true"
      volumes:
        - name: dshm
          emptyDir:
            medium: Memory
            sizeLimit: 8Gi

---
apiVersion: v1
kind: Service
metadata:
  name: dighum-fastpitch-tts-lb
spec:
  type: LoadBalancer
  selector:
    app: dighum-fastpitch-tts
  ports:
  - protocol: TCP
    name: http
    port: 80
    targetPort: 9000
  - protocol: TCP
    name: grpc
    port: 50051
    targetPort: 50051

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: dighum-audio2face-3d
spec:
  replicas: 1
  selector:
    matchLabels:
      app: dighum-audio2face-3d
  template:
    metadata:
      labels:
        app: dighum-audio2face-3d
    spec:
      hostNetwork: true
      containers:
      - name: dighum-audio2face-3d
        image: nvcr.io/nim/nvidia/audio2face-3d:1.2
        securityContext:
          privileged: true
        ports:
        - containerPort: 9000
          name: http
        - containerPort: 50051
          name: grpc
        resources:
          requests:
            nvidia.com/gpu: 2
          limits:
            nvidia.com/gpu: 2
        volumeMounts:
          - name: dshm
            mountPath: /dev/shm
        env:
        - name: NGC_API_KEY
          valueFrom:
            secretKeyRef:
              name: ngc-api-key
              key: NGC_API_KEY
        # - name: NIM_MANIFEST_PROFILE
        #   value: "default"
        - name: NIM_DISABLE_MODEL_DOWNLOAD
          value: "true"
        livenessProbe:
          httpGet:
            path: /v1/health/ready
            port: 9000
          initialDelaySeconds: 10
          timeoutSeconds: 20
          periodSeconds: 10
          failureThreshold: 100
      imagePullSecrets:
      - name: secret-nvcr
      nodeSelector:
        cloud.google.com/gke-gpu: "true"
      volumes:
        - name: dshm
          emptyDir:
            medium: Memory
            sizeLimit: 16Gi

---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: dighum-maxine-audio2face-2d
spec:
  replicas: 1
  selector:
    matchLabels:
      app: dighum-maxine-audio2face-2d
  template:
    metadata:
      labels:
        app: dighum-maxine-audio2face-2d
    spec:
      containers:
      - name: dighum-maxine-audio2face-2d
        image: nvcr.io/nim/nvidia/maxine-audio2face-2d:latest
        ports:
        - containerPort: 8000
          name: http
        - containerPort: 8001
          name: ws
        resources:
          requests:
            nvidia.com/gpu: 1
          limits:
            nvidia.com/gpu: 1
        volumeMounts:
          - name: dshm
            mountPath: /dev/shm
        env:
        - name: NGC_API_KEY
          valueFrom:
            secretKeyRef:
              name: ngc-api-key
              key: NGC_API_KEY
        - name: NIM_HTTP_API_PORT
          value: "8000"
        livenessProbe:
          httpGet:
            path: /v1/health/ready
            port: 8000
          initialDelaySeconds: 10
          timeoutSeconds: 20
          periodSeconds: 10
          failureThreshold: 100
      imagePullSecrets:
      - name: secret-nvcr
      nodeSelector:
        cloud.google.com/gke-gpu: "true"
      volumes:
        - name: dshm
          emptyDir:
            medium: Memory
            sizeLimit: 16Gi

---
apiVersion: v1
kind: Service
metadata:
  name: dighum-maxine-audio2face-2d-lb
spec:
  type: LoadBalancer
  selector:
    app: dighum-maxine-audio2face-2d
  ports:
  - protocol: TCP
    name: http
    port: 80
    targetPort: 8000
  - protocol: TCP
    name: ws
    port: 8001
    targetPort: 8001