Forem: BBCoder

EcoQuest - From Concrete to Canopy with Gemini AI

BBCoder — Sun, 19 Apr 2026 19:12:07 +0000

This is a submission for Weekend Challenge: Earth Day Edition

What I Built

Most environmental apps are just glorified spreadsheets. They ask you to manually log your "carbon footprint" or read articles about distant melting glaciers. They feel like homework. EcoQuest is different because it starts with where you actually live: your neighborhood.

EcoQuest is an AI-powered "Restoration Engine" that turns your local street into a playable quest. It uses vision LLM to analyze photos of your immediate surroundings, identifies ecological "dead zones" (like concrete-heavy parking lots or barren sidewalks), and generates a high-stakes mission to bring nature back.

The goal isn't just to "be greener"—it's to transform yours and your neighbors' physical reality. By completing real-world tasks—from planting micro-forests to installing permeable water-catchment systems—you earn Seeds, unlock Guardian Themes, and contribute to a global, real-time leaderboard of restored hectares.

Demo
Live Demo: https://earth-day-hackathon-2.vercel.app/

Code
GitHub Repository: https://github.com/catcodey/earth-day-hackathon_2

How I Built It

The technical challenge was moving beyond simple "image recognition" into "Visual Reasoning." I didn't just want the app to see a tree; I wanted it to notice the absence of a tree and understand the civic implications.

1) The Vision Engine (Google Gemini)

The "brain" of EcoQuest is Gemini 1.5 Flash. I used Gemini 1.5 Flash not just for captioning, but for structural extraction. On every upload, the app sends the neighborhood image with a complex system prompt. Instead of returning a caption, Gemini returns a structured JSON payload containing:

The Earthiness Index: A score based on greenery density and sustainable infrastructure.
The Gap Analysis: A bulleted report explaining exactly why a neighborhood is "starved" for nature.
Actionable Missions: 3 distinct, actionable custom missions that are specifically designed to improve the earthiness index of your neighborhood.

Once a mission is complete, Guardians upload a "Finished" image. This "Before & After" record serves as the proof of restoration and unlocks the full reward bounty.

2) The Guardian Community & Rewards

Environmentalism is a team sport.

Neighborhood Spotlight: A real-time feed where Guardians from around the world post their transformations. You can "Support" other projects to boost their visibility and foster civic pride.
The Point Loop: Finishing missions earns you Seeds.
The Eco Shop: Seeds are used here to purchase Guardian Themes (like Midnight Forest or Royal Waves) and status badges. This gamification layer ensures that the more you restore the physical world, the more you can personalize your digital presence.

3) Global Stats

Transforms your effort into localized science metrics:

Carbon Offset: Estimated CO₂ sequestered by converting gray concrete to green space.
Water Restoration: Rainwater volume absorbed into the ground via permeable surfaces.
Heat Mitigation: Localized temperature reduction achieved by expanding the urban canopy.
Biodiversity Resilience: Progress toward creating micro-habitats for native pollinators.

4) The Dynamic Theme Engine

Advice is better when it looks good. I implemented a custom Theme & Effect System. When users spend their Seeds (earned via real actions), the entire UI morphs.
Aesthetic Invariants: I built a state-driven CSS bridge where "Equipping" a theme like Midnight Forest or Volcanic Earth updates the primary color tokens, shadows, and borders across the entire app.

5) Celebrating Progress

I used canvas-confetti and motion (Framer Motion) to make "checking a box" feel visceral. Completing a mission doesn't just bump a number—it triggers a visual celebration and updates your Global Impact metrics. These metrics (Carbon offset, Heat mitigation) are calculated dynamically based on the specific missions Gemini generates, moving the impact from "vague idea" to "specific data."

Workflow

Scan: Upload a photo of your neighborhood to the AI Earthiness Analyzer.
Analyze: Gemini 1.5 Flash audits the image and assigns an Earthiness Index.
Select: Choose one of three actionable missions tailored to your block’s needs.
Act: Complete real-world sub-tasks using the persistent mission checklist.
Verify: Upload a "Restored" photo to document the physical transformation.
Quantify: The app calculates CO₂ and water impact based on your success.
Reward: Earn Seeds to unlock exclusive themes and badges in the Eco Shop.
Inspire: Share your Before/After results with the global Guardian community.

Technical Stack

React 19 + Vite: For the blazing-fast SPA scaffolding.
Tailwind CSS 4.0: Powering the "Geometric Balance" design theme.
Google Gemini SDK: For the core visual analysis and mission generation.
React Markdown: Custom-styled to ensure AI reports are high-legibility and beautiful.

Prize Categories
Best Use of Google Gemini
EcoQuest relies entirely on Gemini's multi-modal capabilities. Not only does it "see" the neighborhood, but it also projects a better version of it. The 2x "Mission Reroll" feature showcases Gemini's ability to provide creative variety within a fixed set of constraints, ensuring that no two Guardians have the exact same quest.

The planet doesn't need more spectators; it needs more Guardians. EcoQuest is my attempt to make that transition addictive.

TransitFlow- a productivity app for long distance commuters

BBCoder — Sun, 01 Mar 2026 19:24:45 +0000

This is a submission for the DEV Weekend Challenge: Community

The Community

In today's world, we are drowning in information but starving for wisdom. Every morning, I wake up to the same cycle: 20 unread emails, a 10-item to-do list that only exists as a fog in my head, and a dozen "must-read" tabs I'll never actually open.

Then comes the commute . Every day, I lose close to three hours to the road. If you saw me on the bus, you'd see someone sleeping, listening to music, or staring aimlessly out the window.

When I get to my workplace, I hear the buzz: colleagues discussing the latest Al breakthroughs, current affairs, and new tech shifts. I want to be part of that conversation. I have a burning desire to upskill and stay informed, but on a 3-hour commute, productivity feels impossible. If I try to read, the constant bumping and sudden brakes make me motion-sick. If I try to type,the vibrating screen gives me a headache. It's physically draining to stare down at a small screen when the world is moving outside.

Even when I do manage to skim a thread or listen to a podcast, I'm fighting "Commuter Brain"- a state of mind that makes it nearly impossible to focus and 80% of the time i end up forgetting what i learned just a couple mins ago.

TransitFlow is a one stop solution that turns your commute time into guilt-free, productive, enjoyable experience of growth and learning.

What I Built

App Link
https://ai.studio/apps/affd8a5e-31a7-44d2-bb4e-3ad72e4493fa

TransitFlow is an AI powered productivity suite designed to be your Cognitive Co-pilot.

Core Features
🧠 AI Microlearning: Master any topic in 5-20 minute interactive sessions. The AI doesn't just lecture; it summarizes, asks checking questions, and ends with a quiz to ensure you stay on your toes and actually learn something.
📰 Bite-Sized News: No clickbait, no infinite scrolling, and no distracting YouTube links. Just the top 7 global or tech stories, summarized by AI and ready to be read or listened to.
📧 Inbox - Unread: Connect your Gmail and get a high-level audio summary of your unread emails. Instead of reading 20 threads, you listen to one 60-second brief.
🎙️ Multi-Note Task Maker: Speak your scattered thoughts, and Gemini turns them into organized, actionable project briefs. I implemented a "Sticky Note" system where you can store up to 5 voice-generated notes, with the ability to manually edit or delete them.
🏆 Productivity Gamification: To encourage a balanced workflow, users earn points for every session. Using all four tools in a single day unlocks a "Productivity Bonus," rewarding you for being well-rounded.

App Demo Images

The Workflow:

Home: Greeted by a clean, minimalist interface and an inspiring quote.
Dashboard: The "Productivity Hub" where you track your points and access the four core tools.
Interaction: Whether it's learning about Quantum Physics or summarizing your morning emails, the experience is driven by high-quality AI speech and clean, brutalist-inspired UI.

Code

The project is built as a full-stack application with a React frontend and an Express backend to handle secure OAuth flows.

Key Technical Highlights:

Gemini 3.0 Flash: Powers the reasoning behind the microlearning sessions and the task summarization.
Gemini 2.5 Flash TTS: Provides the "Zephyr" voice that narrates your news and email summaries.
Google OAuth 2.0: A secure backend implementation using google apis and cookie-session to manage Gmail access without exposing tokens to the client.

How I Built It

The Stack

Frontend: React 18, Vite, Tailwind CSS (v4), and Framer Motion for those smooth "app-like" transitions.
Backend: Express.js running on Node.js, serving as a proxy for the Gmail API and handling the OAuth callback.
AI Engine: @google/genai SDK. I used a mix of models: gemini-3-flash-preview for speed and reasoning, and gemini-2.5-flash-preview-tts for high-fidelity speech generation.
Icons: lucide-react for a consistent, professional aesthetic.

The "Aha!" Moment: The Task Maker Logic
One of the most satisfying parts was building the voice-to-brief pipeline. I used the Web Speech API for real-time transcription, then piped that raw text into a Gemini prompt that specifically looks for "actionable items."
code
TypeScript
// The prompt that turns "rambling" into "results"
const prompt = Organize the following spoken thoughts into a clean, actionable project brief with bullet points. ONLY include information from the transcript. Transcript: "${transcript}";

Challenges Overcome

The biggest hurdle was the Gmail OAuth flow within an iframe-based preview environment. Standard redirects don't work well in sandboxed frames. I implemented a popup-based flow with window.postMessage communication, ensuring that once the user grants access in the popup, the main app instantly refreshes its state to show the unread summaries.

Why TransitFlow?

We don't need more apps that steal our attention. We need apps that give us our time back. TransitFlow is my attempt at building a "quiet" tool—one that helps you process the world, learn something new, and get back to what matters.

App link

https://ai.studio/apps/affd8a5e-31a7-44d2-bb4e-3ad72e4493fa

Built with ❤️ for the DEV Community Weekend Challenge.

Submitted By:

Email: bhavna18182003@gmail.com
BBCoder

[Boost]

BBCoder — Thu, 26 Jun 2025 04:20:58 +0000

The Homework Genie: Just Email It!

BBCoder — Mon, 09 Jun 2025 06:59:49 +0000

This is a submission for the Postmark Challenge: Inbox Innovators.

What I Built

I built a "Homework Helper Bot" that transforms email into a powerful AI assistant. It receives math questions via Postmark email, processes them with a local AI (Ollama), generates a detailed PDF solution, and instantly emails it back to the sender, all powered by Postmark.

Demo

Code Repository

Github link: https://github.com/catcodey/postmark-hackathon
postmark url: 57b64f01309581e8ec5ba1bff617115c@inbound.postmarkapp.com

How I Built It

My bot is built with Python, using Flask for webhook handling. It leverages Postmark for both Inbound Email (receiving questions) and Outbound Email (sending PDF solutions), showcasing Postmark's robust APIs.

For AI processing, I integrated Ollama to run a local Large Language Model (phi3:mini), keeping intelligence on my machine. fpdf2 was used for PDF generation, which involved overcoming challenges like Unicode font support and precise text layout within the PDF. The postmarker Python library facilitated reliable outbound email with attachments. ngrok was essential for exposing my local server to Postmark's webhooks.

My experience with Postmark was very positive; its webhook and API functionalities are incredibly effective for building innovative email-driven applications.

Video Link: https://drive.google.com/file/d/1Inh4eR3AkEDREVywIXi4_5fTS-zZN-wM/view?usp=sharing

User name :bhavna_b_baa952ae51dac930 (BBCoder)
Email : bhavna18182003@gmail.com

AuraFeed - Real-time AI News Sentiment Analyzer

BBCoder — Mon, 26 May 2025 07:00:53 +0000

This is a submission for the Bright Data AI Web Access Hackathon

What I Built

This AI agent tackles information overload by automating news collection and analysis. It fetches real-time articles, scrapes their content, then uses AI (like Ollama for sentiment analysis and embeddings for context) to enrich the data. All this structured, intelligent news is stored in a PostgreSQL database and easily viewed through a Streamlit dashboard, transforming a torrent of information into organized, personalized insights.

Demo

code link: https://github.com/catcodey/agent-hackathon
output: https://github.com/catcodey/agent-hackathon/blob/main/Screenshot%202025-05-26%20at%2012.22.38.png

How I Used Bright Data's Infrastructure

I used brightData's web scraper API to gather necessary data from BBC news.I was also able to monitor progress of data collection and store the snapshot data in Postgre for further processing.

User name :bhavna_b_baa952ae51dac930 (BBCoder)
Email : bhavna18182003@gmail.com