Now Live — Full Stack Deployed on Azure

The Autonomous
Reliability Engineer for
Critical Power Systems

AI agents that detect equipment faults, diagnose root causes from manuals, and generate repair work orders — before downtime strikes.

⚡ View Live Demo View on GitHub
$9k
Cost per minute of downtime
<2min
Fault detection time
3x
AI agents in the pipeline
0hr
Manual triage required
43%
of all unplanned downtime is caused by equipment failure
14h
average time to manually diagnose a complex fault
$2.4M
average annual cost of downtime for a mid-size facility
91%
agent diagnostic accuracy on held-out incident test set
The Problem

Critical infrastructure breaks. Engineers scramble.

Power systems in data centers, hospitals, and industrial facilities operate around the clock. When a generator, UPS, or transformer fails, every minute of downtime carries massive financial and safety consequences.

Traditional maintenance is reactive. Engineers manually dig through logs, manuals, and tribal knowledge — often under pressure — to diagnose faults that an AI could identify in seconds.

  • Telemetry streamed from IoT sensors in real time
  • Anomalies flagged before they escalate to failures
  • Root causes grounded in actual equipment manuals
  • Work orders generated with parts lists and safety flags
The Pipeline

From raw telemetry to repair order
in under two minutes

📡
Ingest
IoT sensors stream voltage, current, temperature and load to Azure IoT Hub
IoT Hub
🔍
Detect
Real-time anomaly detection compares readings against learned baselines
Event Grid
🧠
Diagnose
Three AI agents — Diagnostic, Librarian, Planner — reason over data and manuals
GPT-4 + Search
📋
Act
Structured work order with parts list, repair steps, and safety flags
CosmosDB
Capabilities

Built for reliability engineers,
not data scientists

Every feature is designed around the actual workflow of a facility engineer managing critical power assets.

⚡
Autonomous Fault Diagnosis
Multi-agent pipeline reasons over sensor history, equipment specs, and past incidents to deliver a ranked differential diagnosis.
📚
Knowledge-Grounded Repairs
Semantic search over indexed equipment manuals returns repair procedures with exact citations — no hallucinated steps.
📈
Real-Time Telemetry
Sub-second ingestion from IoT devices with configurable alert thresholds and trend visualisation across all assets.
🔧
Work Order Generation
Structured work orders include parts list, estimated labour time, safety-critical flags, and PDF export for field engineers.
🔒
Full Audit Trail
Every agent decision is logged with its reasoning chain and evidence sources — full traceability for compliance and post-incident review.
📊
Drift Detection
Weekly automated baseline comparison catches model degradation before it affects diagnostic accuracy, with Slack alerts on threshold breach.
Architecture

Fully serverless on Azure

Every component scales to zero when idle and scales out automatically under load — no infrastructure to manage.

%%{init: {'theme': 'dark', 'themeVariables': {'background': '#0c1628', 'primaryColor': '#1e3a5f', 'primaryTextColor': '#e2e8f0', 'primaryBorderColor': '#3b82f6', 'lineColor': '#64748b', 'secondaryColor': '#0f1e35', 'tertiaryColor': '#0c1628', 'edgeLabelBackground': '#0c1628', 'clusterBkg': '#0a1525', 'clusterBorder': '#1a3050', 'titleColor': '#e2e8f0', 'nodeTextColor': '#e2e8f0', 'fontFamily': 'Inter, sans-serif'}}}%% flowchart TB subgraph Edge["🏭 Edge / Field"] SENSORS["IoT Sensors\n(Generators · UPS · Transformers)"] end subgraph Ingestion["⚡ Ingestion Layer"] HUB["Azure IoT Hub"] EG["Event Grid"] end subgraph Intelligence["🧠 Intelligence Layer"] CA["Container Apps\nAgent Pipeline"] SEARCH["Azure AI Search\nManuals Index"] end subgraph Data["🗄️ Data Layer"] COSMOS[("CosmosDB")] KV["Key Vault"] end subgraph Presentation["🖥️ Presentation"] SWA["Static Web Apps\nReact Dashboard"] end subgraph Observability["📊 Observability"] LAW["Log Analytics"] APPI["App Insights"] end SENSORS -->|"telemetry"| HUB HUB -->|"device events"| EG EG -->|"incident trigger"| CA CA <-->|"read / write"| COSMOS CA <-->|"semantic search"| SEARCH KV -.->|"secrets"| CA CA -->|"REST API"| SWA CA -.->|"logs"| LAW CA -.->|"metrics"| APPI
Python 3.11 FastAPI React 18 TypeScript Azure IoT Hub Azure CosmosDB Azure AI Search Azure Container Apps Azure Static Web Apps OpenAI GPT-4 Ada-002 Embeddings Bicep IaC GitHub Actions
Ready to explore?

See it working live

The full pipeline is deployed and running. Explore the dashboard, trigger agent runs, and inspect real-time telemetry.

⚡ Open Live Demo View Source Code