Architecture screen

Composable services for fleet telemetry, intervention workflows, and risk APIs.

This screen maps how NADIR moves from ingestion to inference to distribution with strict evidence lineage, tenant isolation, and low-latency API delivery to downstream simulation and OEM workflows.

Ingestion control plane

Provider connectors, schema normalization, and temporal quality checks.

Signed ingest sessions for Samsara and Geotab streams.
Clock reconciliation and missing-frame compensation windows.
Fleet and route scope controls for multi-tenant operations.

Inference mesh

Per-vehicle confidence estimation and persistent drift scoring.

Residual decomposition for orientation and vibration drift.
Threshold policies with route and weather priors.
Dispatch triggers for intervention workflow automation.

Distribution fabric

Evidence exports and dynamic risk APIs from a shared graph.

Claims-ready timeline and closure artifact generation.
Segment-level risk vectors for scenario generation.
REST and SDK surfaces with scoped credentials.

Service boundaries

Primary production services and technical outputs.

Service	Input	Output
connector-gateway	provider telemetry	canonical signal stream
confidence-engine	signal stream + context	drift and confidence events
workflow-orchestrator	risk events	intervention state timeline
risk-segment-api	aggregated evidence graph	route-level risk payloads

Interface examples

from nadir import FleetClient

client = FleetClient(api_key="NADIR_API_KEY")
service_graph = client.architecture.services.list(fleet_id="fleet_mw_01")
print(service_graph.nodes)

import { FleetClient } from "@nadir/sdk";

const client = new FleetClient({ apiKey: process.env.NADIR_API_KEY! });
const matrix = await client.architecture.dependencies({ fleetId: "fleet_mw_01" });
console.log(matrix.edges);

GET /v1/architecture/services?fleet_id=fleet_mw_01
Authorization: Bearer NADIR_API_KEY

{ "nodes": 28, "edges": 114, "version": "2026.03" }