This API provides a system of record for configuration changes across AI deployments. It supports:

• Schema evolution
• Role-based access control
• Audit trail of all changes
• Backwards compatibility

All requests require a bearer token specific to your role:

• Enterprise Infrastructure: enterprise-token
• LOB IT: lob-token
• Application Teams: app-token

Refer to the get_current_role function in LoggingAPI.py for token-to-role mapping.

Enterprise teams manage core infrastructure configurations. This path allows updating the enterprise schema, defining standards like data classification and alerting.

Adding GPU Configuration Example: This curl command demonstrates adding a GPU configuration to the enterprise schema.

curl -X POST http://localhost:8000/api/v1/config/schema \
  -H "Authorization: Bearer enterprise-token" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "gpu-inference",
    "targetMetric": "gpu_utilization",
    "dataClassification": "Restricted",
    "quickAlertHeuristic": {
      "threshold": 0.85,
      "window": "5m"
    },
    "gpu_config": {
      "type": "A100",
      "memory": "80GB"
    },
    "reason": "Production scale-up for high-throughput inference"
  }'

{
  "status": "success",
  "message": "Enterprise schema updated",
  "updated_schema": {
    "type": "object",
    "required": ["name", "targetMetric", "dataClassification", "quickAlertHeuristic"],
    "properties": {
      "name": {"type": "string"},
      "targetMetric": {"type": "string"},
      "dataClassification": {
        "type": "string",
        "enum": ["Public", "Internal", "Confidential", "Restricted"]
      },
      "quickAlertHeuristic": {
        "type": "object",
        "required": ["threshold", "window"],
        "properties": {
          "threshold": {"type": "number"},
          "window": {"type": "string"}
        }
      },
      "author": {"type": "string"},
      "reason": {"type": "string"},
      "timestamp": {"type": "string", "format": "date-time"},
      "gpu_config": {                  # New field added
        "type": "object",
        "properties": {
          "type": {"type": "string"},
          "memory": {"type": "string"}
        }
      }
    },
    "additionalProperties": true
  }
}

LOB IT teams manage runtime configurations and model validation parameters.

Adding Operational Metric Example:

curl -X POST http://localhost:8000/api/v1/runtime/schema \
  -H "Authorization: Bearer lob-token" \
  -H "Content-Type: application/json" \
  -d '{
    "modelVersion": "v2.1.0",
    "validationParameters": {
      "minBatchSize": 32,
      "maxLatencyMs": 100
    },
    "operational_metric": "inference_throughput",
    "reason": "Adding throughput monitoring for batch processing"
  }'

(Successful responses will be similar to the enterprise update, showing "Runtime schema updated" and the updated_schema for "runtime" including the new operational_metric field.)

Application teams manage custom integrations and business metrics.

Adding Custom Metrics Example:

curl -X POST http://localhost:8000/api/v1/integrations/schema \
  -H "Authorization: Bearer app-token" \
  -H "Content-Type: application/json" \
  -d '{
    "customThresholds": {
      "accuracy": 0.95,
      "latency_p99": 250
    },
    "businessMetric": "revenue",
    "callbackUrl": "https://app-endpoint/callback",
    "reason": "Adding latency monitoring for SLA compliance"
  }'

(Successful responses will show "Integration schema updated" and the updated_schema for "integration" including the new callbackUrl field.)

View the history of changes for any schema type (enterprise, runtime, or integration). This is vital for compliance and tracking configuration evolution.

Example Request:

curl -X GET http://localhost:8000/api/v1/audit/enterprise \
  -H "Authorization: Bearer enterprise-token"```

<details>
<summary><strong>📋 Example Response</strong></summary>

```json
{
  "changes": [
    {
      "timestamp": "2024-02-13T14:30:00Z",
      "schema_type": "enterprise",
      "author": "enterprise-enterprise",
      "change": {
        "name": "gpu-inference",
        "targetMetric": "gpu_utilization",
        "dataClassification": "Restricted",
        "quickAlertHeuristic": {
          "threshold": 0.85,
          "window": "5m"
        },
        "gpu_config": {
          "type": "A100",
          "memory": "80GB"
        },
        "reason": "Production scale-up for high-throughput inference",
        "author": "enterprise-enterprise",
        "timestamp": "2024-02-13T14:30:00Z"
      }
    }
    // ... other historical changes for 'enterprise' schema
  ]
}

1. Required Fields (must be present in the request body for the respective schema type, unless already defined and not being changed):
   • Enterprise: name, targetMetric, dataClassification, quickAlertHeuristic
   • LOB IT: modelVersion, validationParameters
   • App Teams: customThresholds, businessMetric
2. All Changes Require:
   • reason: A string explaining the purpose of the change (optional for GET).
   • Appropriate role authorization (validated via token).
   • The request body must validate against the current base schema structure for required fields and their types.
3. New Fields:
   • If additionalProperties is true in the base schema (which it is for all defined schemas), new fields can be added.
   • The API infers the type of new fields (e.g., string, number, object) based on the provided value.
   • These new fields are then incorporated into the schema definition for future validations.
   • New fields cannot override the type or structure of existing, defined fields in base_schemas.

{
  "detail": "Invalid token"
}```
</details>
<details>
<summary><strong>📋 Validation Error (400 Bad Request)</strong></summary>

```json
{
  "detail": "[ErrorDetail(message=\"'reason' is a required property\", ...)]"
}

(Actual jsonschema error message might be more verbose)