Five things I’ve built, what I actually did on each, and what came of it. Two are live and I’m still in them. One I wound down. The rest shipped and I moved on.
LEXI Platform — regulated trade and payments, built now (2025, Co-founder and CEO)
I co-founded LEXI out of GIFT City and run it as CEO. I took its subsidiary, LexiMoney, through IFSCA’s Payment Service Provider process to in-principle approval. The call: ship documentation and compliance first, payments later — the part everyone wants to build is also the part where being early and wrong costs you a licence. Documentation and compliance are live in private beta. Payments are on the roadmap. The PSP approval is in-principle, not final.
MoneeFlo — cross-border payments where the banks don’t go (2023, Co-founder)
I co-founded MoneeFlo, an RBI-licensed forex company, and own the product end to end. The call: build for a physical-plus-digital distribution model, since in tier-2 India the channel partner is the trust layer. Outward remittances went from zero to a peak of $37M in annualised TPV, serving around 2,500 students a year through roughly 400 onboarded channel partners.
Airtel Digital — AI into a twenty-year-old stack (2022, Principal PM)
I sat on customer calls and visited Airtel Centres, and took what customers were saying to leadership in their own words. Only then did I spec the AI that industrialised it—intent classification, agent self-training, moderation. The call: build it inside the old stack rather than beside it. It shipped into production in a stack twenty years old. Not a pilot.
Reliance Jio — an enterprise blockchain from a whiteboard (2018, Product lead, blockchain)
I started it from a whiteboard: the architecture, the approach, the teams. I owned the product for the fintech use cases built on it and left leading four PMs, three architects and around seventy engineers. The patents came out of this work—ten granted US patents, two IEEE papers. The call: most of what people wanted to put on a blockchain didn’t need one. A platform that benchmarked at 35,000 transactions per second, with real fintech products running on it.
Kirana Truck — the one I wound down (2020, Co-founder)
A weekly grocery service on wheels for housing societies during COVID. The call: go weekly and route-based instead of on-demand — made the unit economics work at small scale, capped how big it could get. At its peak it reached 6,500 households a week and sold to 2,500 of them—38% weekly penetration, and profitable. We wound it down anyway; it was capital intensive and seasonal, and no amount of product work was going to change either.
Four questions I keep coming back to
Almost nothing above failed or succeeded on the technology. Before I commit to building something, I ask: who has the power to insist, and will they actually spend it? What does the person I’m forcing to change get out of it? Who here is powerful enough to refuse? And, strip out the technology — would a boring version deliver the same result? The fourth is the one that gets skipped, and it’s the one that saved the most money.