
This is a submission for Weekend Challenge: Earth Day Edition
What I Built
I built EarthLens AI — an intelligent, multimodal environmental analysis platform that acts as a pocket climate scientist for anyone with a smartphone.
The platform has four core features:
- EcoVision — Upload or snap a live photo of any environment (river, forest, urban area, factory). Google Gemini analyzes it and returns an Eco Score, biodiversity assessment, and sustainability evaluation with actionable insights.
- EcoChat — A real-time streaming AI sustainability assistant that answers climate questions and contextualizes your uploaded images.
- Carbon Compass — An interactive calculator that evaluates your personal carbon footprint across transport and lifestyle categories, with AI-powered reduction tips.
- Earth Pulse — A live dashboard of key global environmental indicators.
All results can be downloaded as a PDF report or shared natively via WhatsApp, Email, and other apps directly from your device.
Demo
🌍 Live App (Google Cloud Run): https://earthlens-1006999380256.us-central1.run.app
The app is a fully installable Progressive Web App (PWA) — on mobile, tap "Add to Home Screen" for a native app experience!
Code
parulmalhotraiitk
/
earthlens
EarthLens AI is a multimodal environmental intelligence platform powered by Google Gemini. Upload photos of habitats, analyze daily carbon impacts, and chat about global sustainability through the lens of AI.
🌍 EarthLens AI
An intelligent, multimodal environmental analysis platform powered by Google Gemini.
🔗 Live Demo: earthlens-1006999380256.us-central1.run.app
🚀 Overview
EarthLens AI brings the complex world of climate science down to a personal, actionable level by leveraging the powerful multimodal capabilities of the Google Gemini API. Designed to act as a pocket environmental scientist, the application allows users to assess ecological health, track their personal footprint, and learn about sustainability effortlessly.
✨ Key Features
| Feature | Description |
|---|---|
| 📸 EcoVision | Upload or snap live photos of any environment — rivers, forests, cities, factories — and get an instant AI-powered Eco Score, Key Findings, and Recommended Actions |
| 💬 EcoChat | A conversational climate assistant powered by Gemini that can answer any question about sustainability, biodiversity, or climate change |
| 👣 Carbon Compass | A 4-step lifestyle questionnaire that calculates your personal carbon footprint and generates a tailored Gemini action plan |
| 🌍 Earth Pulse | Live global climate |
How I Built It
Frontend: Vanilla HTML5, CSS3, and JavaScript — no heavy frameworks. I built a premium glassmorphic dark-mode UI with CSS Grid, smooth micro-animations, and a fully responsive mobile-first layout down to 320px screens.
Backend: A lightweight Node.js + Express.js proxy server sits between the browser and Google Gemini. This keeps the API key completely server-side (stored in Cloud Run environment variables) and applies an in-memory rate limiter (30 req/min per IP) to prevent abuse.
Google Gemini Integration:
-
EcoVision uses the multimodal
generateContentendpoint — base64-encoded images are sent alongside structured prompts that force consistent scoring output. -
EcoChat uses
streamGenerateContentwith Server-Sent Events (SSE) piped directly from Gemini → Node → browser, creating a real-time typewriter streaming effect. - The model used throughout is
gemini-3-flash-previewfor speed and multimodal capability.
Deployment: The entire app is containerized with Docker and deployed serverlessly on Google Cloud Run with automatic scaling.
Prize Categories
Best Use of Google Gemini — EarthLens AI is fundamentally built around Gemini's multimodal capabilities. The entire platform — from live image-based ecological scoring in EcoVision, to real-time streaming conversations in EcoChat — would not be possible without Gemini's ability to reason across both images and text simultaneously.
United States
NORTH AMERICA
Related News
UCP Variant Data: The #1 Reason Agent Checkouts Fail
7h ago
Amazon Employees Are 'Tokenmaxxing' Due To Pressure To Use AI Tools
21h ago
How Braze’s CTO is rethinking engineering for the agentic area
11h ago

Décryptage technique : Comment builder un téléchargeur de vidéos Reddit performant (DASH, HLS & WebAssembly)
17h ago
How AI Reduced Manual Driver Verification by 75% — Operations Case Study. Part 2
4h ago
