This is a submission for the Presage SmartSpectra SDK Challenge
What I Built
Hyperion Core: Building a Transparent Seismic Research Dashboard
Earthquake information is important—and the way we present it matters. With Hyperion Core, I’m exploring how a software prototype can bring together public earthquake catalog data, interactive visualizations, and small training demos while being clear about what is real and what is simulated.
Hyperion Core is a Japan-focused, bilingual research and training dashboard. It fetches recent earthquake events from the USGS past-hour catalog, filters them to a broad Japan-region area, and displays details such as magnitude, depth, location, and event time. The dashboard also includes a Japan-centered globe, a prefecture directory, preparedness resources, an offline-capable app shell, and local peer-to-peer messaging demos.
The technology behind it
The main dashboard is built with Next.js, React, and TypeScript. JavaScript, HTML, and CSS support browser interactions and interface styling.
The repository also contains bounded examples and starter modules in Rust, Zig, C++20, Julia, Python, Haskell, Go, Java, R, PHP, and Swift. These explore ideas such as frame parsing, signal processing, scientific modeling, and inference adapters. They are examples and scaffolds—not a collection of production-ready services.
What models does it use?
There are two distinct model-related features:
-
Gemini Live API: An optional, user-started English/Japanese voice conversation feature. The configured default model is
gemini-3.8-live. It supports conversational interaction; it is not an earthquake prediction model or an official alert channel. - A logistic regression demo: A browser-based training lab generates a synthetic dataset and trains a small classifier. It demonstrates a machine-learning workflow, but the data is artificial and the results say nothing about real earthquake risk.
Hyperion Core does not include a validated or pretrained seismic prediction model. Its visualized signals and hardware telemetry are illustrative, not measurements.
Why be explicit about the limits?
Earthquake and emergency information can influence important decisions. This project is not connected to field sensors, Japan’s official Earthquake Early Warning system, emergency dispatch, sirens, or infrastructure controls. Its visuals are demonstrations, and the USGS catalog lists detected events—it is not a real-time warning feed.
Please use JMA and local authorities for official information and emergency guidance. Hyperion Core is a research prototype for exploring software and communicating data boundaries, not a tool for operational or emergency decisions.
Live demo: https://hyperion-core-secure-edge.aniruddhadhibar5.workers.dev
webdev #nextjs #typescript #react #machinelearning #earthquake
I based the language and model details on the repository: Gemini Live is configurable, the logistic regression lab uses generated data, and no validated seismic predictor is included.
Code
<!-- Share a link to your public code repository. Make sure it includes setup instructions. --> https://github.com/aniruddhadhibar5-bot/ApexTectonic
Top comments (0)