a

I design for difficult decisions.

I came to product through graphic design, fashion, and a denim label I co-founded. That route still shapes how I work: I look for the structure beneath the interface, reduce what competes for attention, and make difficult decisions easier to see.

A career made of different materials

  • 2013–2017 · NIFT Chennai, Lee CooperVisual foundations

    Composition is not decoration. It decides what gets seen first, what waits and what disappears. A beautiful direction still has to survive production and the hands that make it. What stayed: hierarchy, typography, rhythm, respect for constraints, and the difference between a concept and a thing that ships.

  • 2019–2021 · AirtelConsumer scale

    At scale, one unclear choice becomes millions of moments of friction. What stayed: behavioural evidence, support that answers what someone meant rather than what they typed, and the knowledge that small interface choices compound.

  • 2021–now · Sirion.aiEnterprise AI

    AI can answer. A person still has to judge the evidence, consequence and next move. What stayed: decision systems, provenance, actions you can undo, human checkpoints, and interfaces that stay steady when the model is unsure.

Experience

  • Senior UX Designer Sirion.ai Enterprise contract management, used by global legal teams.
  • UX Designer Bharti Airtel Airtel Thanks, one of India’s largest consumer apps.
  • UI/UX Designer Think Matra
  • Senior Creative UI Designer Great Place to Study
  • Co-founder and Creative Director Zhero Denim
  • Graphic and Product Designer Lee Cooper

Bachelor of Design, NIFT Chennai · Executive Development Programme in Strategy and Leadership, XLRI Delhi-NCR (2025–26) · Design Thinking, Stanford Graduate School of Business (online).

This idea was picked for UX India 2023

My concept for AI-assisted contract redlining and review was selected for a Lightning Talk at UX India 2023 in Bengaluru. It pointed reviewers to the risky parts of a contract, suggested fixes, and drafted replacement clauses. The initial concept and the talk were mine; I didn’t work on the version that later shipped.

Auto Redline as it later shipped in Sirion, from Sirion’s product film.

How I work now

HypothesisWhat do I believe? BuildMake it real TestWhat happens? IterateChange it LEARNINGLOOP
  1. HypothesisI sit with the people doing the work and write down what I think is true, before I draw anything.
  2. BuildA working prototype in HTML, built with AI tools, so people react to the real thing instead of a picture of it.
  3. TestThe people who do the work try it on their own tasks. What they do counts for more than what they say.
  4. IterateChange the solution, not the slide. Then start the next round.
Amrit at a whiteboard mapping smart data capture for real-world contracts: upload and OCR, extraction, key attributes, an AI engine, reasoning that explains why a value was extracted, human-in-the-loop review with confidence scores, repository, integrations and continuous learning, all radiating from a single starting point.
Most of the hard calls are settled on a wall like this, weeks before anyone opens a design tool.
Amrit with four colleagues from the Sirion product team, in the office.
With the Sirion product team. Most of what is on this site was argued out with people like these before it was drawn.

Some of it I learn at home. My daughter rarely explains where an experience failed. She pauses, tries something else, or leaves. It left me a sharper eye for where confidence breaks, and respect for what people do instead of what they say.