01

Projects

Ganesha Tracker

2026

Community map · Bengaluru

→ 34 pandals on the map

The problem. During Ganesh Chaturthi, Bengaluru fills with pandals and there is no single place to find them. People ask on Instagram and Reddit and get a name, with no address, timing, or photo.

What I built. A map-first site for Ganesh Chaturthi 2026. Pandals are plotted across the city, with photos, neighbourhood search, WhatsApp share, and a form to add one while you are out.

  • 34 pandals listed, with 137 community favourites
  • 4,622 page views during the festival
  • Filter by community, temple, or apartment, and search by neighbourhood
  • React
  • Leaflet
  • FastAPI
  • MongoDB
  • Product Discovery
  • User Stories

Get Me Home

2026

Pay-to-rank · Fiction

→ $1 to list · $10k to land

The problem. Creators want to be seen, and buying the same attention with ads is expensive. A bare leaderboard gives no one a reason to look.

What I built. A pay-to-rank board with Commander Sam’s rescue as the skin. Paste a link, take a spot, and every listing adds fuel that moves his ship toward Earth. A payment buys a ranked listing. The rescue is fiction.

  • $1 lists a URL. Pay $1 over the seat above to overtake it
  • The ship on the first fold moves as the board fills, toward a $10,000 landing
  • Live at getmehome.lol, with hosted checkout and a public board
  • React
  • Supabase
  • Dodo
  • Feature Prioritization
  • UI/UX Design

DhundoJob

2026

AI outreach · Solo build

→ 3 hours to 5 minutes

The problem. Researching a company, finding the right person, and writing a cold email that sounds like you was a three-hour grind. Every time.

What I built. A solo product — PM, design, and engineering — that does the research, the discovery, and the personalization, then leaves a five-minute flow. It runs in the browser: Gemini-grounded research, email discovery, and multi-persona drafts, with local state and JSON export. No backend.

  • 60+ personalized emails sent, about 80% opened
  • Research, discovery, and outreach collapsed from 3 hours to 5 minutes
  • Still evolving against live use
  • Next.js
  • TypeScript
  • Tailwind
  • Gemini
  • Vercel
  • Product Discovery
  • AI Product Strategy

OTFNA

2026

Demo food order · India

→ The ritual, none of the bill

The problem. Late-night hunger still triggers the urge to order food online. The taps that close that loop — browse, cart, checkout — used to end in a real charge.

What I built. OTFNA (Only The Food Never Arrives) is a phone-first demo of that loop. Ten fictional restaurants, cuisine filters, a cart, checkout, and fake tracking. No signup, no payment, no rider. History keeps rupees not spent, streaks, and badges on this device.

  • 10 demo restaurants, with search, cuisine filters, and sort
  • 14 home-cook recipes, plus a history of demo orders and money not spent
  • No account. Orders, streaks, and badges stay in local storage
  • Next.js
  • Tailwind
  • PWA
  • Product Discovery
  • UI/UX Design

Get in Flow

2026

Focus · Pomodoro

→ Focus time, tied to a task

The problem. The timer, the to-do list, and the noise all live in different apps. You finish a session and still cannot say where the time went.

What I built. One screen: a 25-minute Pomodoro, the task you are working on, and an ambient soundscape. Completed sessions log against that task. Everything stays in the browser, with JSON export.

  • Focus and break timer, with time recorded on the task you picked
  • Rain, forest, ocean, and white noise
  • Tasks and session history survive a reload and export as JSON
  • Next.js
  • Tailwind
  • Web Audio
  • Feature Prioritization
  • UI/UX Design

Orangelist

2022

Productivity · Flutter

→ Live web demo

The problem. Most to-do apps are either too heavy to open or too plain to want to.

What I built. A small cross-platform to-do product in Flutter — playful enough to open, light enough to finish the list.

  • Shipped as a Flutter app with a live web demo
  • Open source, built for everyday task planning
  • Flutter
  • UI/UX Design
  • User Stories