Core Services

Auth, AI Processing, Event Handling

Role in the Project

These services manage core functionalities:

• Authentication (Auth0, Keycloak): User identity, role-based access control (RBAC), and secure sessions.
• AI Processing: Handling real-time object detection requests and rule-based automation.
• Event Handling: Managing asynchronous AI tasks and external integrations.

Strengths & Weaknesses

Strengths:

• Decouples authentication and AI tasks for modularity
• Event-driven architecture improves responsiveness and scaling

Weaknesses:

• Requires well-defined API rate limits to prevent abuse
• Needs robust monitoring to avoid processing bottlenecks

Available Technologies & Comparison

• Authentication: Auth0 (managed), Keycloak (self-hosted), Firebase Auth (serverless, Google-dependent)
• AI Processing: TensorRT (optimized for NVIDIA GPUs), ONNX Runtime (cross-platform), PyTorch Serve (flexibility for custom models)
• Event Handling: Kafka (high throughput, complex setup), MQTT (lightweight for IoT), RabbitMQ (balanced approach)

Chosen Approach

• Keycloak for authentication (self-hosted, customizable, supports OAuth2, JWT, SAML)
• PyTorch Serve for AI model inference (flexibility and GPU acceleration)
• Kafka for high-throughput event streaming (enables real-time automation)

Example of an event handling flow:

{
  "event": "object_detected",
  "timestamp": 1714500000,
  "data": {
    "imageId": "12345",
    "objects": [
      { "label": "Person", "confidence": 0.98 },
      { "label": "Car", "confidence": 0.95 }
    ]
  }
}