Uplauz

The idea

Live shows are full of moments that never get captured — the song someone is dying to hear, the tip they'd happily send, the dedication they'd shout if they could. Uplauz turns a live audience into a participating, paying one: scan a QR to join, request songs, send UPI tips, drop dedications and polls, and watch it all surface on the big screen behind the artist.

It's a live-event revenue and engagement OS for artists and venues.

My role

I'm the technical cofounder and the person who builds it. Local git history shows 322 of 323 commits authored by me — from the first Next.js commit through payments, the realtime backend, media/livestream, telemetry, and AI song detection. Three of us on the team; I own all of engineering.

Frontend

Next.js / React — audience, artist, and big-screen surfaces

Backend

Convex realtime — requests, tips, polls, presence

Payments

UPI flows for tips and support

Media

Cloudflare-backed livestream + media infra

Intelligence

LLM-assisted song detection from live transcripts

A piece I'm proud of: live song detection

During a show, the system listens to a realtime transcript and uses an LLM classifier — with a lock policy and fallback logic — to detect which song is playing, so requests and the screen stay in sync with the set. It shipped with tests and an architecture doc. The strongest pilot for it ran with the band Agnee.

Traction

Built from zero to a live product with real shows. Secured an AIC Manipur grant (Atal Incubation Centre) and ran artist/venue pilots — including work with Agnee, Indofunk, and others.

designed & built solo

0→1

Manipur grant

AIC

real shows run

live

What I took from it

Founder mode is full-stack in the literal sense: the database schema and the on-stage UX are the same problem. Shipping something people use while it's happening live is a brutal, wonderful constraint — there's no "deploy and check tomorrow." It either works during the encore or it doesn't.