name: anchor-nvidia-gb10

services:
  mongodb:
    image: mongo:8.0@sha256:02a0cc7939f5ed38f30f9bc714ef5f682d49baf9350c54acf302ce833087fe8a
    environment:
      MONGO_INITDB_ROOT_USERNAME: anchor
      MONGO_INITDB_ROOT_PASSWORD: ${MONGO_PASSWORD:-anchor-gb10-local-only}
      MONGO_INITDB_DATABASE: anchor
    volumes:
      - mongo-data:/data/db
    tmpfs:
      - /tmp:size=128m,mode=1777
    security_opt:
      - no-new-privileges:true
    restart: unless-stopped
    healthcheck:
      test:
        - CMD-SHELL
        - >-
          mongosh --quiet --host 127.0.0.1
          --username $$MONGO_INITDB_ROOT_USERNAME
          --password $$MONGO_INITDB_ROOT_PASSWORD
          --authenticationDatabase admin
          --eval "quit(db.adminCommand('ping').ok ? 0 : 2)"
      interval: 10s
      timeout: 5s
      retries: 20
      start_period: 20s

  mongo-express:
    image: mongo-express:1.0.2-20@sha256:30fc6333779340217c6a2bc0bc3cd53aa46028ab8fc6caef41bd00bf0faacf53
    profiles: [inspect]
    ports:
      - "127.0.0.1:${MONGO_EXPRESS_PORT:-8081}:8081"
    environment:
      ME_CONFIG_MONGODB_URL: mongodb://anchor:${MONGO_PASSWORD:-anchor-gb10-local-only}@mongodb:27017/?authSource=admin
      ME_CONFIG_BASICAUTH_ENABLED: "true"
      ME_CONFIG_BASICAUTH_USERNAME: ${MONGO_EXPRESS_USERNAME:-anchor-inspector}
      ME_CONFIG_BASICAUTH_PASSWORD: ${MONGO_EXPRESS_PASSWORD:-anchor-gb10-inspect-local-only}
    depends_on:
      mongodb:
        condition: service_healthy
    read_only: true
    tmpfs:
      - /tmp:size=32m,mode=1777
    security_opt:
      - no-new-privileges:true
    cap_drop:
      - ALL
    restart: unless-stopped


  nemotron:
    image: ${NVIDIA_VLLM_IMAGE:-nvcr.io/nvidia/vllm:26.05.post1-py3}
    pull_policy: if_not_present
    entrypoint: ["/bin/bash", "-c"]
    command:
      - |
        vllm serve /models/nemotron \
          --served-model-name nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4 \
          --host 0.0.0.0 \
          --port 8000 \
          --max-model-len ${NEMOTRON_MAX_MODEL_LEN:-8192} \
          --gpu-memory-utilization 0.35 \
          --max-num-seqs 8 \
          --trust-remote-code \
          --enable-auto-tool-choice \
          --tool-call-parser qwen3_coder \
          --reasoning-parser nemotron_v3 \
          --moe-backend flashinfer_cutlass \
          --async-scheduling \
          --kv-cache-dtype fp8
    environment:
      FLASHINFER_DISABLE_VERSION_CHECK: "1"
    ports:
      - "127.0.0.1:${NEMOTRON_PORT:-18000}:8000"
    volumes:
      - ${NEMOTRON_MODEL_DIR:?Set NEMOTRON_MODEL_DIR}:/models/nemotron:ro
      - nemotron-cache:/root/.cache/vllm
    shm_size: 16gb
    ulimits:
      nofile: 2048
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    restart: unless-stopped
    healthcheck:
      test: ["CMD-SHELL", "curl -f http://127.0.0.1:8000/health || exit 1"]
      interval: 20s
      timeout: 10s
      retries: 90
      start_period: 120s

  digital-twin:
    image: ${ANCHOR_APP_IMAGE:-anchor-nvidia-gb10:latest}
    build:
      context: app
      dockerfile: Dockerfile.csm-cuda
      additional_contexts:
        csm_wheels: ${CARELINE_CSM_WHEELHOUSE:?Set CARELINE_CSM_WHEELHOUSE}
    ports:
      - "${CARELINE_BIND_ADDRESS:-0.0.0.0}:${CARELINE_PORT:-8100}:8100"
    environment:
      CARELINE_LLM_BASE_URL: http://nemotron:8000/v1
      CARELINE_LLM_MODEL: nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4
      CARELINE_LLM_STRONG_BASE_URL: http://nemotron:8000/v1
      CARELINE_LLM_STRONG_MODEL: nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-NVFP4
      CARELINE_LLM_MAX_TOKENS: "120"
      CARELINE_REASONING_BUDGET_TURN: "0"
      CARELINE_REASONING_BUDGET_EXTRACT: "0"
      CARELINE_CLONE_BACKEND: cuda
      CARELINE_CSM_MODEL_DIR: /models/sesame-csm-1b
      CARELINE_SELF_REF_AUDIO: /run/careline/self_ref.wav
      CARELINE_SELF_REF_TEXT: /run/careline/self_ref.txt
      CARELINE_STT_BACKEND: local-whisper
      CARELINE_STT_MODEL_DIR: /models/whisper-tiny-en
      CARELINE_TTS_BACKEND: nim
      CARELINE_MONGO_URI: mongodb://anchor:${MONGO_PASSWORD:-anchor-gb10-local-only}@mongodb:27017/anchor?authSource=admin
      CARELINE_MONGO_DB: anchor
      ANCHOR_DEMO_ACCESS_KEY: ${ANCHOR_DEMO_ACCESS_KEY:-}
      CARELINE_OPENCLAW_WAKE_URL: ${CARELINE_OPENCLAW_WAKE_URL:-}
      CARELINE_OPENCLAW_HOOK_TOKEN: ${CARELINE_OPENCLAW_HOOK_TOKEN:-}
    volumes:
      - ${CARELINE_CSM_MODEL_DIR:?Set CARELINE_CSM_MODEL_DIR}:/models/sesame-csm-1b:ro
      - ${CARELINE_STT_MODEL_DIR:?Set CARELINE_STT_MODEL_DIR}:/models/whisper-tiny-en:ro
      - ${CARELINE_SELF_REF_AUDIO:?Set CARELINE_SELF_REF_AUDIO}:/run/careline/self_ref.wav:ro
      - ${CARELINE_SELF_REF_TEXT:?Set CARELINE_SELF_REF_TEXT}:/run/careline/self_ref.txt:ro
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    init: true
    read_only: true
    tmpfs:
      - /tmp:size=1g,mode=1777
    security_opt:
      - no-new-privileges:true
    cap_drop:
      - ALL
    restart: unless-stopped
    depends_on:
      nemotron:
        condition: service_healthy
      mongodb:
        condition: service_healthy
    healthcheck:
      test: ["CMD", "python", "-c", "import json, urllib.request; data=json.load(urllib.request.urlopen('http://127.0.0.1:8100/api/status')); assert data['ready']"]
      interval: 10s
      timeout: 5s
      retries: 30
      start_period: 30s

  share-proxy:
    image: cloudflare/cloudflared:2026.8.2@sha256:0aa26e284f05e6c77ae375b8c9c11d9eb6a448fb7bcd8d40f31cb6176189eb38
    profiles: [share]
    command: tunnel --no-autoupdate --url http://digital-twin:8100
    depends_on:
      digital-twin:
        condition: service_healthy
    read_only: true
    tmpfs:
      - /tmp:size=16m,mode=1777
    security_opt:
      - no-new-privileges:true
    cap_drop:
      - ALL
    restart: unless-stopped

volumes:
  mongo-data:
  nemotron-cache: