NexusCity

An AI-driven smart city dashboard with live IoT sensor streaming, dynamic disaster response protocols, and automated emergency SMS alerts.

Comment

NexusCity

AI-Driven Proactive Smart City Incident Response VoltHacks 2026 Submission

Inspiration

Megacities face compounding crises daily. A simple traffic accident during a storm can cascade into gridlock, delayed ambulances, and isolated power outages. Existing city management dashboards are purely reactive—they show operators that a crisis is happening, but leave the complex, multi-sector coordination (traffic, medical, energy) to slow human manual processes.

We realized that by fusing IoT edge sensor networks with Generative AI, we could transform urban incident response from a reactive monitoring state to a proactive, automated dispatch system.

What it does

NexusCity is an AI orchestrator that sits on top of a city's IoT infrastructure. It acts as the central brain for urban crisis management:

  1. IoT Data Ingestion: It ingests real-time telemetry via WebSockets from edge devices (traffic flow sensors, microgrid power meters, and air quality trackers).
  2. Generative AI Consultant: When a multi-variable emergency strikes, our LLM-powered engine instantly matches the scenario with municipal Standard Operating Procedures (SOPs).
  3. Automated Cross-Sector Dispatch: Instead of isolated responses, NexusCity synchronizes actions like rerouting traffic lights, instructing microgrids, and dispatching SMS alerts via Twilio.

🛠️ How we built it

  • Cloud Infrastructure: Hosted on Cloudflare Pages for lightning-fast edge delivery, with a Node.js/Express backend handling WebSocket streams.
  • IoT Integration: We engineered a live WebSocket streaming service that simulates 7 physical sensor nodes reporting real-time telemetry and anomaly detection.
  • AI Engine: Powered by Google Gemini 1.5 Flash, with Featherless.ai seamlessly integrated as a robust LLM fallback mechanism to ensure 100% uptime during crises.
  • Frontend Dashboard: A React + Vite Digital Twin interface providing real-time data visualization.

Challenges we ran into

Integrating diverse live IoT data streams (traffic vs. energy) into a unified state for the frontend was challenging. Furthermore, ensuring the Generative AI engine didn't "hallucinate" emergency responses required strict guardrails and prompt engineering based on real-world SOPs.

Accomplishments that we're proud of

  • Successfully bridging the gap between abstract LLM reasoning and real-time IoT data streams.
  • Building a robust LLM fallback chain (Gemini → Featherless.ai) to guarantee system reliability.
  • Translating complex physical infrastructure logic into an intuitive, zero-latency React dashboard.

What's next for NexusCity

  • Direct Edge AI Deployment: Pushing Small Language Models (SLMs) directly to edge IoT gateways.
  • Telecom Integration: Expanding our Twilio integration to support massive Cell Broadcast Systems (CBS) for city-wide evacuation routing.

Built With

Share this project:

Updates

Submission history