Architecture Overview
The Omnierax Intelligence Stack is a layered architecture that separates data acquisition, ontology and fusion, decision orchestration, and operational presentation into independently scalable subsystems. Each layer exposes a typed interface to the layer above and is operated as an independent service with its own deployment cadence, observability surface, and security boundary.
This document is the canonical architecture reference for technical evaluators, deployment architects, and integration engineers. It is read alongside the deployment-model-specific guides (Air-Gap, AWS GovCloud, Azure Government, Private Cloud, Edge) which apply the architecture to specific operational environments.
1. Layer Topology
The stack comprises four primary layers and two cross-cutting subsystems:
- Acquisition Layer — source adapters, schema mapping, ingest validation
- Ontology & Fusion Layer — entity resolution, knowledge graph, temporal ontology
- Decision Layer — agent orchestration, model inference, human authority enforcement
- Presentation Layer — operational interfaces, mission workspaces, dashboards
- Security & Governance (cross-cutting) — zero-trust policy, audit ledger, KMS
- Observability (cross-cutting) — telemetry, traces, model performance monitoring
2. Reference Configuration
omnierax:
deployment: air-gap
region: sovereign-1
layers:
acquisition:
adapters: [opc-ua, mqtt, kafka, syslog, stix-taxii]
ingest_rps_target: 250000
fusion:
graph_engine: omnierax-kg
partitions: 256
er_threshold: 0.9985
decision:
orchestrator: omnierax-orx
latency_target_p99_ms: 25
human_authority: enforced
presentation:
mission_workspaces: enabled3. Versioning & Compatibility
Documentation is maintained for the current release and the two preceding major releases. A version selector at the top of every page allows operators to view documentation for their specific deployment version.