Project Omni: Climate Resilience Analytics

Global food-security analytics: FAOSTAT crop yields correlated with temperature anomalies, forecast to 2050.

Live demo: https://project-omni-frontend-zrnk.onrender.com · API: https://project-omni-backend-zrnk.onrender.com

Project Omni is an end-to-end data intelligence platform that analyzes, predicts, and visualizes the impact of climate change on global agriculture. It correlates decades of FAOSTAT crop-yield data with temperature anomalies to score regional resilience and project yields per country and crop.

Features

• Interactive global heatmap — Leaflet-based globe visualizing crop resilience scores across every continent.
• Micro-linear forecasting — 17,000+ independent per-country, per-crop linear regression models projecting yields to 2050.
• Functional data pipeline — a Haskell ingestion engine that cleans and formats raw FAOSTAT data with ADT-based domain modeling and Maybe/Either error handling.
• Resilience scoring — detrending algorithms that isolate climate shocks from technological progress into a 0–10 "True Resilience" score.
• Premium dashboard — glassmorphic UI built with Next.js, Framer Motion, and Tailwind CSS.

Architecture

flowchart LR
    A[Raw FAOSTAT CSV] --> B[Haskell pipeline<br/>parse, clean, aggregate]
    B --> C[cleaned_crop_data.csv]
    C --> D[Python backend<br/>stdlib HTTP server + OLS ensemble]
    D --> E[JSON REST API<br/>/api/data, /api/predict, /api/resilience]
    E --> F[Next.js frontend<br/>globe, charts, dashboard]
    G[Gemini API<br/>optional market simulation] --> D

Loading

Tech Stack

Layer	Technology
Data pipeline	Haskell (Stack), pure functional transforms
Backend	Python 3 standard library (http.server), dependency-free OLS, optional Gemini enrichment
Frontend	Next.js 16, React 19, TypeScript, Tailwind CSS 4, Framer Motion, Leaflet, Recharts
Deployment	Docker, Render (render.yaml)

Note on the backend: the API is implemented with Python's standard library (http.server / BaseHTTPRequestHandler) — it does not use FastAPI or Flask. The only third-party runtime dependencies are pycountry (ISO-3 lookup) and the optional google-generativeai SDK for market simulation. Heavier models (XGBoost, SARIMAX, Prophet) are loaded opportunistically when installed and gracefully skipped otherwise.

Quick Start (Docker)

docker compose up --build

• Dashboard: http://localhost:3000
• API: http://localhost:8000/api/metadata

Quick Start (local)

1. Pipeline (optional — cleaned data ships in the repo):

   cd haskell-pipeline && stack run

2. Backend:

   cd python-backend
   python -m venv venv && venv/Scripts/activate   # or source venv/bin/activate
   pip install -r requirements.txt
   python main.py

3. Frontend:

   cd frontend
   npm ci
   npm run dev

Open http://localhost:3000.

Environment Variables

Variable	Service	Required	Description
GEMINI_API_KEY	backend	No	Enables Gemini-powered market simulation; the API runs without it.
DATA_PATH	backend	No	Absolute path to cleaned_crop_data.csv; defaults to python-backend/data/cleaned_crop_data.csv.
PORT	backend	No	Listen port; defaults to 8000 (Render sets it automatically).
NEXT_PUBLIC_API_URL	frontend	No	Backend base URL, e.g. https://project-omni-backend-zrnk.onrender.com. Inlined into the client bundle at build time (see frontend/Dockerfile); defaults to http://localhost:8000.

Copy .env.example (create your own locally — never commit real values) or set them in your platform's dashboard.

Testing

pip install -r python-backend/requirements.txt pytest
pytest python-backend/tests -q

The suite covers the dependency-free OLS model (fit/predict/confidence bands), ISO-3 country resolution, registry integrity, and the bundled dataset shape.

Deployment

Both services deploy to Render from render.yaml (Blueprint):

• project-omni-backend — Docker runtime, serves the API.
• project-omni-frontend — Docker runtime, serves the dashboard.

Set GEMINI_API_KEY and NEXT_PUBLIC_API_URL in the Render dashboard after the first deploy.

Roadmap

• Replace the stdlib HTTP server with a typed ASGI framework if the API surface grows.
• Move the cleaned dataset to object storage with lazy download to slim the repo.
• Add frontend unit tests (Vitest) alongside the backend suite.
• Cache Gemini simulation responses to cut latency and cost.

Contributing

1. Fork and create a feature branch (git checkout -b feat/my-change).
2. Run ruff check python-backend --select E4,E7,E9,F and pytest python-backend/tests -q before committing.
3. Open a PR; CI must be green before merge.

License

MIT © Akash Naickar. Created for the Advanced Agentic Coding Hackathon.