peregrine/compose.yml

# compose.yml — Peregrine by Circuit Forge LLC
# Streamlit (app service) removed — Vue+FastAPI is the only frontend (#104)
services:

  api:
    image: ghcr.io/circuitforgellc/peregrine:latest
    build:
      context: .
      dockerfile: Dockerfile
    command: >
      bash -c "uvicorn dev_api:app --host 0.0.0.0 --port 8601"      
    volumes:
      - ./config:/app/config
      - ./data:/app/data
      - ${DOCS_DIR:-~/Documents/JobSearch}:/docs
    environment:
      - STAGING_DB=/app/data/staging.db
      - DOCS_DIR=/docs
      - ANTHROPIC_API_KEY=${ANTHROPIC_API_KEY:-}
      - OPENAI_COMPAT_URL=${OPENAI_COMPAT_URL:-}
      - OPENAI_COMPAT_KEY=${OPENAI_COMPAT_KEY:-}
      - PEREGRINE_GPU_COUNT=${PEREGRINE_GPU_COUNT:-0}
      - PEREGRINE_GPU_NAMES=${PEREGRINE_GPU_NAMES:-}
      - GPU_SERVER_URL=${GPU_SERVER_URL:-${CF_ORCH_URL:-http://host.docker.internal:7700}}
      - CF_ORCH_URL=${CF_ORCH_URL:-${GPU_SERVER_URL:-http://host.docker.internal:7700}}
      - CF_APP_NAME=peregrine
      - MNEMO_HOST=${MNEMO_HOST:-mnemo}
      - MNEMO_PORT=${MNEMO_PORT:-8080}
      - PYTHONUNBUFFERED=1
    extra_hosts:
      - "host.docker.internal:host-gateway"
    restart: unless-stopped

  web:
    image: ghcr.io/circuitforgellc/peregrine-web:latest
    build:
      context: .
      dockerfile: docker/web/Dockerfile
    ports:
      - "${VUE_PORT:-8506}:80"
    depends_on:
      - api
    restart: unless-stopped

  searxng:
    image: searxng/searxng:latest
    ports:
      - "${SEARXNG_PORT:-8888}:8080"
    volumes:
      - ./docker/searxng:/etc/searxng:ro
    healthcheck:
      test: ["CMD", "wget", "-q", "--spider", "http://localhost:8080/"]
      interval: 10s
      timeout: 5s
      retries: 3
    restart: unless-stopped

  ollama:
    image: ollama/ollama:latest
    ports:
      - "${OLLAMA_PORT:-11434}:11434"
    volumes:
      - ${OLLAMA_MODELS_DIR:-~/models/ollama}:/root/.ollama
      - ./docker/ollama/entrypoint.sh:/entrypoint.sh
    environment:
      - OLLAMA_MODELS=/root/.ollama
      - DEFAULT_OLLAMA_MODEL=${OLLAMA_DEFAULT_MODEL:-llama3.2:3b}
    entrypoint: ["/bin/bash", "/entrypoint.sh"]
    profiles: [cpu, single-gpu, dual-gpu-ollama, dual-gpu-vllm, dual-gpu-mixed]
    restart: unless-stopped

  ollama_research:
    image: ollama/ollama:latest
    ports:
      - "${OLLAMA_RESEARCH_PORT:-11435}:11434"
    volumes:
      - ${OLLAMA_MODELS_DIR:-~/models/ollama}:/root/.ollama
      - ./docker/ollama/entrypoint.sh:/entrypoint.sh
    environment:
      - OLLAMA_MODELS=/root/.ollama
      - DEFAULT_OLLAMA_MODEL=${OLLAMA_RESEARCH_MODEL:-llama3.2:3b}
    entrypoint: ["/bin/bash", "/entrypoint.sh"]
    profiles: [dual-gpu-ollama, dual-gpu-mixed]
    restart: unless-stopped

  vision:
    build:
      context: .
      dockerfile: scripts/vision_service/Dockerfile
    ports:
      - "${VISION_PORT:-8002}:8002"
    environment:
      - VISION_MODEL=${VISION_MODEL:-vikhyatk/moondream2}
      - VISION_REVISION=${VISION_REVISION:-2025-01-09}
    profiles: [single-gpu, dual-gpu-ollama, dual-gpu-vllm, dual-gpu-mixed]
    restart: unless-stopped

  cf-orch-agent:
    build:
      context: ..
      dockerfile: peregrine/Dockerfile.cfcore
    command: ["/bin/sh", "/app/docker/cf-orch-agent/start.sh"]
    ports:
      - "${CF_ORCH_AGENT_PORT:-7701}:7701"
    environment:
      - CF_ORCH_COORDINATOR_URL=${CF_ORCH_COORDINATOR_URL:-http://host.docker.internal:7700}
      - CF_ORCH_NODE_ID=${CF_ORCH_NODE_ID:-peregrine}
      - CF_ORCH_AGENT_PORT=${CF_ORCH_AGENT_PORT:-7701}
      - CF_ORCH_ADVERTISE_HOST=${CF_ORCH_ADVERTISE_HOST:-}
      - PYTHONUNBUFFERED=1
    extra_hosts:
      - "host.docker.internal:host-gateway"
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
    profiles: [single-gpu, dual-gpu-ollama, dual-gpu-vllm, dual-gpu-mixed]
    restart: unless-stopped

  mnemo:
    image: ghcr.io/zaydmulani09/mnemo:latest
    ports:
      - "${MNEMO_PORT:-8080}:8080"
    volumes:
      - mnemo-data:/data
    environment:
      - MNEMO_DB_PATH=/data/mnemo.db
      - MNEMO_LLM_PROVIDER=${MNEMO_LLM_PROVIDER:-ollama}
      - MNEMO_LLM_BASE_URL=${MNEMO_LLM_BASE_URL:-http://ollama:11434/v1}
      - MNEMO_LLM_API_KEY=${MNEMO_LLM_API_KEY:-ollama}
      - MNEMO_LLM_MODEL=${MNEMO_LLM_MODEL:-llama3.2:3b}
    depends_on:
      - ollama
    healthcheck:
      test: ["CMD", "wget", "-q", "--spider", "http://localhost:8080/health"]
      interval: 15s
      timeout: 5s
      retries: 3
    profiles: [memory]
    restart: unless-stopped

  finetune:
    build:
      context: .
      dockerfile: Dockerfile.finetune
    volumes:
      - ${DOCS_DIR:-~/Documents/JobSearch}:/docs
      - ${OLLAMA_MODELS_DIR:-~/models/ollama}:/ollama-models
      - ./config:/app/config
    environment:
      - DOCS_DIR=/docs
      - OLLAMA_URL=http://ollama:11434
      - OLLAMA_MODELS_MOUNT=/ollama-models
      - OLLAMA_MODELS_OLLAMA_PATH=/root/.ollama
    profiles: [finetune]
    restart: "no"

volumes:
  mnemo-data: