Science was always a conversation. But now you're in it.
🏆 1st Place — ML & AI Tooling Challenge, DataHacks 2026 (Best Use of Marimo/Sphinx) Built in 36 hours at UC San Diego by Kiruthika Marikumaran & Mallika Dasgupta.
Watson & Crick is an AI-powered environmental health research engine. You ask a plain-English question — Does PM2.5 pollution affect Alzheimer's risk? What does wildfire smoke do to my lungs? — and two AI scientists argue about your data in real time, run real statistics, and hand you a peer-review-grade research report in about 30 seconds.
Watson runs the analysis optimistically. Crick questions every assumption. Together they produce:
- A causal graph showing the pathway from environmental exposure → candidate genes → health outcome (built in D3.js with force-directed layout, edge weights derived from real correlation strength)
- Three ranked hypotheses with biological mechanisms
- A structured 8-section research report
- A confidence score with a full breakdown of every factor and weight
- Audio playback, PDF export, and a community feed where anyone can publish and fork research
- 🔬 Research Engine — Pearson correlations, Random Forest models, Isolation Forest anomaly detection, real p-values
- 🌡️ Live UCSD Campus Safety Dashboard — fuses real Scripps AWN sensor data with EPA PM2.5 and NWS heat index; SAFE/CAUTION/ALERT system with K-Means zone clustering and Random Forest temperature forecasting
- 📓 Marimo Interactive Notebook — exposes the exact statistical engine with live reactive sliders so anyone can reproduce or challenge a result in real time
- 🗺️ San Diego Health Map — solar permit density vs. respiratory health outcomes across neighborhoods
- 💬 Ask the Scientists — follow-up chatbot powered by Gemini 2.5 Flash
| Dataset | Source |
|---|---|
| Air quality sensors | EPA AQS |
| Climate records | NOAA |
| Disease prevalence | CDC PLACES |
| Genetic variants | EMBL-EBI GWAS Catalog |
| Live campus sensors | Scripps Institution AWN @ UCSD |
Frontend: React, Vite, Tailwind CSS, TanStack Router, D3.js, jsPDF
Backend: Python, FastAPI
ML: scikit-learn (Random Forest, Ridge Regression, Isolation Forest, K-Means), SciPy
AI: Gemini 2.5 Flash API, Claude
Notebook: Marimo
Infra: Vercel, Lovable
# Clone the repo
git clone https://github.com/kirustar14/EcoCausal.git
cd EcoCausal
# Backend
cd backend
pip install -r requirements.txt
uvicorn main:app --reload
# Frontend
cd frontend
npm install
npm run devYou'll need a Gemini API key set as GEMINI_API_KEY in your environment.
DataHacks is the annual 36-hour MLH-certified hackathon hosted by the Data Science Student Society (DS3) at UC San Diego.