Vanguard Preparedness

Vanguard is a survival intelligence hub bridging the gap between complex data and human action. Powered by Gemini 3 Pro, it works with Datadog to turn raw telemetry into life-saving clarity.

Comment

Inspiration

I wanted to bridge the gap between high-level engineering data and the everyday person who needs clarity in a crisis. Witnessing how easily the grid can fail, my goal was to create a resilient "Mission Control" that makes complex telemetry accessible to anyone, ensuring that life-saving clarity isn't reserved only for engineers.

What it does

Vanguard acts as a "Mission Control" for the individual, designed to work with both a modern iPad and a 10-year-old Raspberry Pi 2, translating complex environmental telemetry and inventory manifests into human-centric survival strategies.

Vanguard is a dual-layered survival hub that cross-references AI tactical reasoning with real-world hardware vitals. I use Gemini 3 Pro to process inventory manifests and tactical intelligence, while a unified Datadog dashboard monitors critical metrics, including System Readiness, environmental seismic activity, and the physical health of the Pi itself.

The main Vanguard app contains the following:

  • a main dashboard for information on weather temperature, solar, cloud cover, earthquakes and floods.
  • System readiness calculates a percentage score based on real-time environmental data to show if current conditions are safe for operation.
  • Manifest where the user can log any item in their inventory
  • Utilities where the user can send an emergency flash, morse code, use a unit converter and look up common survival phrases in 5 different languages
  • AI Uplink where in online mode the user can speak to Gemini 3 Pro
  • An Intel Map showing the nearest source for resources like food and water
  • Protocols contains offline resources for survival and online resources for government links
  • Training contains three different games where you can practice Reflexes, Cognition and Signals.

The Datadog app contains the following:

  • Detection alerts generated by Gemini for both the iPad app and the Pi
  • The Live API data from the government is fed through Gemini and into Datadog
  • Logs for food and water inventory are stored
  • Crew Readiness logs the gaming metrics and an overall Training Proficiency score

How I built it

I built the user interface as a React and TypeScript application, using Google AI Studio to integrate Gemini 3 Pro for advanced survival logic. On the hardware side, I deployed the Datadog IoT Agent on a Raspberry Pi 2 to stream metrics like CPU temperature, RAM usage, and disk health directly to a custom operations dashboard. I designed the UI using Google AI Studio and prompting to achieve a high end sleek look and feel then manually created the Datadog dashboard using metrics retrieved from AI Studio, the Raspberry Pi 2 and several free API's. The API's I used were: UK Environment Agency (Flood Data), Open-Meteo (Cloud Cover, Solar, Temperature) and USGS Earthquake.

Challenges I ran into

The primary hurdle was the hardware limitation of the Pi 2’s ARMv7 architecture, which required a lightweight approach rather than a standard cloud deployment. I had to manually optimize the Datadog IoT Agent and write custom log-cleaning automation to prevent the SD card from failing, a common "Pi killer" in long-running missions.

Accomplishments that I'm proud of

I am incredibly proud of achieving pro-tier observability on a decade-old device while maintaining a high-contrast, tactical UI that doesn't feel like a "developer tool." Successfully triggering a physical "kill-switch" demo where unplugging the Pi creates an actionable Datadog Incident case was a major technical milestone for me.

What I learned

This project taught me that the "Edge" is just as much about hardware maintenance as it is about AI intelligence. I learned how to manage memory constraints on limited systems and discovered the power of using a "Thinking" model like Gemini 3 Pro to generate a practical interface that can communicate with the Datadog system to create a product that solves real-world problems.

What's next for Vanguard Preparedness

I plan to expand Vanguard’s "Tactical Intel" layer by integrating LoRaWAN support for true off-grid communication between multiple hubs. I also want to refine the AI's "Thinking" telemetry to prioritise even lower-bandwidth responses, ensuring Vanguard remains the ultimate lifeline in the most extreme conditions.

Share this project:

Updates