Eukleídes — Team Cobalt

NASA Space Apps Challenge 2025: “A World Away – Hunting for Exoplanets with AI”

---

🧭 Overview

Eukleídes is an AI-powered web application built by Team Cobalt for the NASA Space Apps Challenge 2025.
Our mission is to streamline exoplanet classification using machine learning and interactive visualizations. Making it easier for researchers and astronomy enthusiasts to identify potentially habitable worlds.

---

🧠 What It Does

• 🪐 Exoplanet Classification — Uses a trained ML model to determine whether a candidate is a confirmed exoplanet.
• 📊 Geometric Visualizations — Transforms prediction results into intuitive, interactive visuals inspired by orbital geometry.
• ⚙️ Two Modes:
  1. Single Exoplanet Analysis — Quick lookup for a single candidate.
  2. Dataset Upload — Bulk classification for research teams.
• 🛰️ Responsive, Accessible Web UI — Optimized for both researchers and public outreach.

---

🧩 How It Works

1. Frontend (Landing App)

   • Built with HTML, CSS, TypeScript, and hosted on Vercel or GitHub Pages
   • Features: video background, smooth-scroll, interactive hero section

2. Backend (API)

   • Built with FastAPI (Python)
   • Receives planetary parameters
   • Runs inference using trained ML model
   • Returns probability and classification result

3. Deployment

   • Frontend deployed via Vercel or GitHub Pages
   • Backend hosted on Render (/predict endpoint)

---

🧰 Tech Stack

Layer	Tools Used
Frontend	HTML, CSS, TypeScript, React, 3JS, NextJS
Backend	Python, FastAPI, scikit-learn, Pandas
AI/ML	RandomForestClassifier trained on NASA Exoplanet Archive
Hosting	Vercel (UI), Render (API)

---

⚗️ Model Input Parameters

The model uses six numerical features per exoplanet:

1. Orbital Period
2. Planet Radius
3. Stellar Effective Temperature
4. Stellar Radius
5. Transit Depth
6. Transit Duration

Frontend (Static Site)

git clone https://github.com/rajeevphysics/cobalt.git
cd lander
npm install
npm run dev
# or open dist/index.html after build