Get to know me
Shipped features that make companies millions every month; won hackathons. I've also killed two of my own products.
They all come from the same skill & Villain's Hat's VRSD methodology came out of doing both.
They all come from the same skill & Villain's Hat's VRSD methodology came out of doing both.
The work
Lived and worked across 4 continents. Designed and shipped products across Finance, Consultancy, Energy, Entertainment, and Aerospace. Led teams, hit KPIs, moved metrics that C-suite, finance and legal teams trust and engineers respect. Also walked away from things that didn't meet my bar.
2 hackathons won, 1 came second. Ryanair's first 24-hour hackathon. Smartbox's AI Hackathon. Smartbox's Integrations Hackathon. Not for the trophies. For the practice of shipping a real thing on a tight deadline, with compliance and accessibility intact.
A specific story
At Ryanair, I designed an AR bag sizer that helped users measure their bags before they flew. It took me 8+ months to get the business case into the backlog. Users loved it. The ML accuracy improved by the day. Numbers tracked perfectly with user feedback and research.
Then it was quietly shelved.
It worked so well it stopped people from making expensive bag upgrades at the airport. Fewer people in the queues. Margins dropped. The metric that paid my salary got worse the moment the feature helped users get what they actually needed.
I learned to notice when a design that's right for the user is wrong for the business, and when it's the other way around. I learned to nail and name the tension instead of pretending it isn't there. I learned that the user the product never wrote for isn't always vulnerable. Sometimes it's the user the product wrote for, served too well.
That tension is what Villain's Hat formalized.
The origin of Villain's Hat
I'm telling you this because it's where the methodology originated from. It's not the only thing I've done but it's one that I've landed on after numerous trial and errors. The killed products are why I wrote the methodology.
This was a really fun one, until it wasn't. Roastfluencer was a Gemini-powered roast generator. It emulated standup-comedian kind of roasts, a person just needed to upload a picture. It was technically clean, the roasts worked, but the guardrails slipped. I could have shipped it, but I put myself in the shoes of others who may use this.
Imagine this: A group of kids in a school corridor using it to gang up on a classmate they've just bullied and taken a photo of. They upload the photo to Roastfluencer, and the roast has them pointing and laughing at that kid. I've been the bullied, I know how much it hurts without AI in the mix. The AI made the cruelty cheaper, faster, and at scale. I decided to kill it before launch.
A platform for people in critical health situations to leave messages for their loved ones upto 10 years into the future. echonote was originally inspired a few years after I lost my mother. I wish I could hear her voice, her laughter, hear her nag even. This was the kind of product I cared about; the kind that, if it worked, would matter the world to me.
I built seven AI guardrail agents. I named each one, gave them a persona with JDs. I tested them in a multitude of scenarios. Then a passing thought changed it all. What if, just what if, someone is trying to blackmail someone and the messages are scheduled in advance, maybe had pictures attached too? Another example is, imagine this. 'I love you' from someone you love (like my mom) vs 'I love you' from a stalker will have very different reactions. One will bring memories and tears, the other will paralyze and make one shiver in angst. I'd hate to think the effect this would have on someone already broken by grief, getting hit again on the anniversary.
The guardrails caught most scenarios. They couldn't catch enough of them, and so I shut it down.
I wouldn't consider closing my own products naivety. It's a quality bar. I know I could work on the AI agents but would that stop the humans from finding workarounds to do evil?
This is where I draw the line. If I won't ship a product I'd be ashamed of, I won't help you ship one either.
What I'm building now
The methodology that came out of 16 years in UX and killing those two products is now applied across everything I do.
The VRSD framework. Free Claude Skill and CustomGPT as a taster, full Investigations tool launching in H2 2026.
Read the methodology →Co-founded with Om Tandon. Product audits for games and enterprise software. A recent smart-hydration App won bronze at A'Design Awards 2026.
Visit UX Reviewer →Long-form conversations with founders. Season 2 is all about founders who almost shipped a villain, and what they did to fix it.
Watch Season 1 on YouTube →Patterns I find in client work, online, and in founder conversations. Written every week.
Read the Substack →
Belgrade, 2026. Talking about probabilistic UX for AI agents when they think back.
Where I am
Living in Belgrade has its own challenges with the language, but they're all dog-lovers here, so it's great to be here with Om and our cocker spaniel Pari. We're in the process of adopting a child. We've lived across 4 continents - India, New Zealand, Ireland, and now Serbia. I'm learning Serbian, and I already speak English, Hindi, Kannada, plus a tiny bit of Spanish, Portuguese, Korean, Chinese and Japanese.
I've wanted to get into Y Combinator for 20 years and I applied last year with a fashion app focused on transitioning shoppers experiences from volume driven to value driven which in turn changes how fast fashion brands slow down, reducing amount of garments that land in our landfills. They said no. The reason I think was I had the business case, I had a prototype, but no Tech co-founder. Villain's Hat is the version I'm building with AI as my Tech co-founder.
If you want to work together, the fastest paths are: try the free Quick Check, apply to the next workshop cohort, or DM me on LinkedIn (links below).
Start with the free Quick Check. Or read the methodology. Or DM me on LinkedIn.