inec/docker-compose.ml.yml at main · munisp/inec · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
version: "3.8"

# INEC ML Infrastructure Stack
# Run with: docker-compose -f docker-compose.ml.yml up

services:
  # Python ML Inference Server (PaddleOCR, InsightFace, CDCN, GNN)
  ml-inference:
    restart: unless-stopped
    build:
      context: ./ml
      dockerfile: Dockerfile
    ports:
      - "8090:8090"
    environment:
      - MODELS_DIR=/app/models
      - PYTHONPATH=/app
    volumes:
      - ml-models:/app/models
    deploy:
      resources:
        limits:
          memory: 4G
          cpus: "4"
    healthcheck:
      test: ["CMD", "python", "-c", "import urllib.request; urllib.request.urlopen('http://localhost:8090/health')"]
      interval: 30s
      timeout: 10s
      retries: 3

  # Rust High-Performance Inference Engine (ONNX Runtime, batch scoring)
  rust-inference:
    restart: unless-stopped
    build:
      context: ./services/inference-engine
    ports:
      - "8091:8091"
    environment:
      - MODELS_DIR=/app/models
      - NEO4J_URI=bolt://neo4j:7687
      - NEO4J_USER=neo4j
      - NEO4J_PASSWORD=${NEO4J_PASSWORD:?NEO4J_PASSWORD must be set}
      - RUST_LOG=info
    volumes:
      - ml-models:/app/models
    depends_on:
      - neo4j
    deploy:
      resources:
        limits:
          memory: 2G
          cpus: "4"
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8091/health"]
      interval: 30s
      timeout: 5s
      retries: 3

  # Neo4j Graph Database (polling unit adjacency, fraud rings)
  neo4j:
    restart: unless-stopped
    image: neo4j:5.17-community
    ports:
      - "127.0.0.1:7474:7474"  # Browser (loopback only)
      - "127.0.0.1:7687:7687"  # Bolt (loopback only)
    environment:
      - NEO4J_AUTH=neo4j/${NEO4J_PASSWORD:?NEO4J_PASSWORD must be set}
      - NEO4J_PLUGINS=["apoc"]
      - NEO4J_server_memory_heap_initial__size=512m
      - NEO4J_server_memory_heap_max__size=1G
    volumes:
      - neo4j-data:/data
    deploy:
      resources:
        limits:
          memory: 2G

  # Model Training (run once to generate model weights)
  train:
    restart: "no"  # one-shot training job (profiles: training); must not auto-restart
    build:
      context: ./ml
      dockerfile: Dockerfile
    command: >
      bash -c "
        python training/anomaly_detection/train.py --output /app/models &&
        python training/liveness_pad/train.py --output /app/models --epochs 10 &&
        python training/gnn_network/train.py --output /app/models --epochs 50 &&
        python training/face_recognition/train.py --action metadata
      "
    volumes:
      - ml-models:/app/models
    profiles:
      - training
    deploy:
      resources:
        limits:
          memory: 4G
          cpus: "4"

volumes:
  ml-models:
    name: inec-ml-models
  neo4j-data:
    name: inec-neo4j-data