Quality, specified!
The Standard
Every run answers to a standard I wrote first, so the output is checked rather than trusted.
I build the systems that design and the teams that run them.
Design Lead at EPAM Systems. I work on how design organisations should operate once AI becomes part of the delivery infrastructure, not simply on how designers use it.
Review, built in!
Self-Review
Every agent checks its own work against it before a person is ever asked to look.
Judgment, kept!
The Call
The agent flags what falls short. Whether that is really a problem is still my call.
Pipeline, shared!
Everyone Runs It
Anyone brings a brief in, the pipeline runs the same way, and they review at the checkpoints.
Gaps, reported!
The Team
The team runs it on their own work, and every gap and better idea comes straight back.
Pipeline, refined!
The Revision
User needs, team notes, run results, business goals: I weigh them, refine the pipeline, it starts again.
To those building design organisations
An Open Letter
Most of what gets called AI-native design is a faster way to make the same artefacts. The drawings arrive sooner. Nothing else changes.
The question I work on is not how designers use AI. It is how a design organisation should operate once AI is part of the delivery infrastructure: how complex design work decomposes into reusable skills, specialised agents, orchestration, validation, and the human decision points that hold the whole thing accountable.
That last part is the part everyone skips. A pipeline that grades its own work and passes is not a pipeline, it is a demo. Something has to be able to stop it, and someone has to be willing to.
I deliberately separate automation from delegation.
- Automate Some problems should simply be automated.
- Delegate Some should be delegated to specialized AI agents.
- Collaborate Some require collaboration between humans and AI.
- Decide And some decisions should remain explicitly human.
Design leadership in an AI-native organization requires knowing the difference.
Avadhesh Kumar Singh
Recognition
Awarded inside the studio, by the people who saw the work land.
Certification
Assessed outside it, against someone else’s bar.
Work
Selected work
- 01 Design as an Agentic Pipeline Building DesignX, an AI-native orchestrator that turns a brief into a defensible, traceable design system of record, and a working app.
- 02 IMF Governance Platform Turning 26 institutional documents into three working platforms in a two-week RFP sprint
- 03 Buy Now, Redeem Later Leading design across RBI's multi-brand platform, and the revenue-model discovery that asked whether Burger King could sell future meal value
- 04 When the Design Was Ready But the Engine Wasn't The honest story of Piplsay, a quantitative research platform that shipped, gathered 15,000 respondents across 14 countries, and got shelved.