Bharath S

AI-first Product Designer in Chennai, IN · bharath.sparky@gmail.com

I design complex systems that feel simple.

I’ve spent five years at Zuper, joining as one of the founding designers of the new platform and growing into an associate lead. Before that: blockchain — NFT marketplaces, exchanges, wallet dashboards. The through-line is complexity; I go where the workflows are heaviest.

My default move is subtraction. Workflow builders, automation logic, CPQ, settings, purchasing — the value lives in the logic, so the interface has to stay out of the way. I win when a workflow gets shorter.

My process is AI-first: MCP servers and n8n automations handle the repetitive parts of design ops, so my time goes to problem definition and system thinking. Evidence over opinion — product analytics, recorded customer calls, and what already exists in the product before inventing anything new.

AI-first designer who does the product-manager half too

Research, analytics, and specs run through agents and MCP servers, so my time goes to problem definition instead of file management.

Customer calls, read at scale

Fathom MCP pulls recorded customer calls straight into my working session. I search and cluster what people actually said before I open Figma — not a summary of a summary.

Analytics before opinions

PostHog MCP answers “do they even use this?” in the same window I’m designing in. I check the number before I argue for the change, not after someone asks.

Specs and notes that write themselves

Notion, Granola, and Obsidian connect through MCP, so meeting notes become requirements and requirements become specs without me retyping any of it.

Prototypes in code, not slides

Claude Code and Conductor let me build the real interaction to test it. Engineers get a working reference instead of a redline, which is why my handoffs are short.

The PM half of the job

Scoping, writing the brief, sizing the problem, chasing the data. I do that work anyway — the agents just make it fast enough to do it and still design.

how I work

Obvious in five seconds

If a dispatcher has to think about the interface, the interface has failed. Clarity beats cleverness, every time.

Reuse before invention

A new pattern is a debt the whole product pays. I exhaust the design system before proposing anything new — I maintain it, so I feel it.

Design for the worst shift

Low battery, direct sunlight, wet gloves, one bar of signal. If it works there, it works everywhere.

experience

Zuper — Associate Lead Product Designer

Dec 2024 — now

Design lead for the roofing vertical. Moved quoting out of spreadsheets and hand-built PDFs and into the product — configure, price and send a proposal without leaving it — then shipped a document builder for contract paperwork and a 0→1 mobile job gallery so site photos hold up as evidence in an insurance claim.

Run design on an agent toolchain: MCP servers into product analytics and recorded customer calls, and prototypes built in code rather than redlined. That is how the product-manager half — scoping, briefs, sizing — gets done without eating the design time.

Maintain the design system, which means every new pattern I approve is a debt I service myself.

Zuper — Senior Product Designer

Dec 2023 — Dec 2024

Rebuilt Settings end to end. The fix was information architecture rather than UI — the map changed far more than the screens did — and the new structure was pressure-tested in usability tests with PM and engineering before any of it shipped.

Shipped the dispatch board, the purchasing module, and the AR aging and invoicing flows.

Zuper — Founding Product Designer

Sep 2021 — Dec 2023

One of the founding designers of the current platform: jobs, scheduling and customer management.

Built the Customer Portal, and the standards set there grew into Zuper’s formal design system.

Blockchain App Factory — Product Designer · Sr. UX Engineer

Oct 2018 — Sep 2021

Trading and custody interfaces — NFT marketplace, exchange and wallet dashboards — where misreading a number costs the user money. Three years of dense, high-stakes data UI before field service.

Upwork — Freelance UI Designer & Developer

Jun 2016 — Aug 2018

Designed and built client work end to end, including a school-management platform. The years spent shipping front-end code are why the handoffs still come out buildable.

outside work

Namma Madras — நம்ம மெட்ராஸ்

A free, independent directory for people arriving in Chennai with very little — food, night shelters, healthcare, work leads, learning spaces and hygiene facilities, in one place.

Most of it works offline after the first load. The people it is for do not have reliable data, and a service finder that needs a connection is not a service finder. It is the same principle as designing for a technician on one bar of signal, taken further.

It says on its own front page that it is not a government app and that listings may be out of date — because someone crossing the city to a shelter that has moved is not an inconvenience, it is a night outdoors. Being trusted mattered more than looking official.

on Google Play

Conductor — Figma Make Athon, March 2026

A DJ deck you play with your hands and a webcam. Wave to move particles and shape the sound, make a fist to strip the vocals out of any track you upload, pinch to draw in mid-air. Sessions record to video, and a shareable code lets someone else join yours.

The interesting constraint was that it has almost no interface. Every control is a gesture, which means the feedback has to carry the whole burden of teaching — you learn what your hands do by seeing the particles answer, not by reading a label.

It is also the clearest evidence I have that I still build. Hand tracking, live audio separation and a particle system in a browser is not a thing you can hand to an engineer as a redline.

try it in a browser

Zuper Glass — A field-service concept for smart glasses

A working prototype that moves Zuper’s field-service flow onto smart glasses — “quick setup, faster inspections, built for the field”, for a roofer whose hands are on the ladder, not a phone.

This one is an experiment on the product I work on rather than away from it — the same field problems I design for on a phone, moved to a surface where both hands stay free. The glass treatment is the argument, not decoration: a heads-up layer has to sit over the real world without hiding it.

try it in a browser

selected work

Open the full portfolio →