Skip to content

About

I kept taking the hardest problem in the room.

Seven years of that led me to design and AI.

I started as a graphic designer at a small edtech startup. Me, a CEO who thought in revenue, and a queue of things that had to look good and work. I moved into UX chasing the second half of that.

In UX I started using AI to push my own thinking rather than skip it. Seven years in, I design products that reason with the person using them.

(01) How I got here

  1. 2019–2022

    Tutelage

    The only designer in the building. I learned to ask what a screen was supposed to sell before I designed it.

  2. 2022–2024

    Tier5

    I built Friender's design system from scratch. Most of what mattered happened before anyone opened Figma.

  3. 2024–2025

    DigiMantra

    Enterprise platforms, where briefs arrived half-written. I spent the first week of a project working out what the client was asking for.

  4. 2025–Present

    Finvasia

    First designer in the building, again. This time on AI products for traders moving their own money, in a domain where half the constraints are regulatory and nobody thinks to tell you.

(02) What I believe

The work I threw away was work I started too early.

  • AI is a thinking partner

    I use it to argue with my first idea. It gives me the three alternatives I would have skipped.

  • Clarity is the job

    I don't open a design tool until I can state the problem in a sentence. The work I've thrown away was work I started too early.

  • Build to learn

    I built UXMantra to find out how an AI product behaves when I own every decision inside it.

(03) Currently

Working at
Finvasia — Senior Product Designer. Building Jumpp and Shoonya.
Building
UXMantra, a UX research partner that shows its sources, in private alpha. whysogood, a hundred-plus free browser tools, live.
Interested in
AI behaviour design, system prompt design, conversation design, AI red teaming.
Open to
Senior design roles at AI-focused companies and labs.

(04) Beyond the work

I spend as much time on how an AI should behave as on how it looks.

That took me into behaviour design, system prompt design, and the problem of what a good AI interaction feels like to sit through.

I write about it on LinkedIn and build experiments on weekends. I also use Claude to pressure-test the same ideas I'm designing around, which I've stopped finding strange.

The full record All work