StarShield - NASA CAD Asteroid Risk Prediction System

A production-ready asteroid risk prediction system built on real NASA JPL Close Approach Data (CAD).

Setup

1. Install Dependencies

pip install -r requirements.txt

2. Process NASA Data (Already Done)

# Convert NASA CAD data to training format
python convert_cad_data.py

3. Train Model (Already Done)

# Train RandomForest model on NASA data
python train_model.py

4. Start API

# Start the FastAPI server
python asteroid_api.py

The API will be available at: http://localhost:8000

5. Setup Frontend and Run

open a new terminal window

cd frontend; npm i
npm run dev

The frontend will be available at: http://localhost:8080

Made by Vidur Shah, Nicolas Asanov, Arya Venkatesan, Abhimanyu Agashe as part of Carolina Data Challenge 2025