Six chapters, so far.

Mechanical engineering at IIT Jodhpur, on the edge of the Thar desert. The classroom taught thermodynamics; the workshop taught something more durable: physical objects do not negotiate. The proving ground was SAE BAJA, where student teams build a single-seat all-terrain vehicle from scratch. First season: I modelled parts, ran the structural analysis, and we placed 13th in India for presentation out of 370+ teams. Next season: I led all 25 of us and took charge of the brakes. You think differently when your classmate is the crash test.

I also found I liked organising people as much as machines: 25 event heads at our technical festival, 40 students on a schools initiative that drew 500+ children. The whole plot, in miniature.

First job: Vizara Technologies, digitising India’s cultural heritage. I started at the bottom of the stack, registering laser-scanned point clouds of ancient monuments and fixing the 3D printers when they jammed at 2am. Four years and three promotions later, I ran product and business development. In between: the largest 3D-printed collection of UNESCO World Heritage Site replicas anywhere, built with five engineers in materials 1.5x stronger at half the cost, shown at the National Museum in New Delhi and the India Pavilion at Dubai Expo 2020, and publicly recognised by India’s Prime Minister. When the pandemic cancelled IIT Jodhpur’s convocation, we staged it in mixed reality and Geoffrey Hinton joined virtually. The workshop years even produced published research, on low-cost natural binders for 3D-printing sand moulds.

The commercial education was just as steep: a business line grown 40% in six months, then eight XR products in eighteen months with revenue up 90%, five institutions on multi-year contracts, and deals across India, the UAE and Singapore.

Vizara taught me product as a craft; needl.ai, an AI knowledge platform, taught me product as a science. I spent a year as an associate product manager letting data argue with my instincts: fifteen A/B tests across 50,000+ users, a signup funnel rebuilt until drop-offs fell by a fifth, and 10,000 user sessions read to find where onboarding actually lost people. Completion went from 40% to 85%. Recurring revenue grew 30%. An AI search filter I shipped hit 15% adoption in its first quarter and moved NPS up eight points.

The year cured me of a common product disease: the belief that taste is enough. Taste proposes. Cohorts decide.

Almost every AI education product starts in English, which quietly excludes hundreds of millions of students. So I founded ShikshaSathi, a Hindi-first AI learning platform for the national curriculum: Llama 3 fine-tuned on textbook content, with an answer-validation framework so the AI was checked before a student relied on it. Ola supported it. A year later came AceItChamp, an adaptive revision and assessment platform piloted in three coaching institutes. Both non-profits, both run beside a demanding day job.

Neither made me a rupee, which was the point. When your users are sixteen and your budget is zero, you learn exactly what is essential and what is theatre.

In late 2023, Instill, a Los Angeles company building an AI-powered Culture Operating System for enterprises, made me its first product hire. The brief was wide: own the product, run the technical programme, figure out the governance. I led eight people across five countries, took the enterprise beta live in three months, and watched early demand become $500K in pilot contracts within six, at 90% retention. We fine-tuned a language model on fifty years of behavioural research and lifted its accuracy by 40%. A rebuilt onboarding lifted weekly active users by 80%.

The work I am proudest of is the least glamorous: the bias audit. If software is going to analyse meetings, someone has to check what it mishears, and for whom. I wrote the methodology, the fairness tests, the thresholds that trigger investigation, and the path to a human when something looks wrong. It is the difference between AI a company will pilot and AI a company will keep.

By 2025 I had shipped enough to know exactly which questions I could not yet answer: pricing a strategy, weighing a market, making a case a board would sign. So I came to the University of Edinburgh Business School and treated the MBA as a working apprenticeship. For Lloyd’s Register I built an AI deployment strategy with a £3.9M annual value case. For Oundle Town Council, a five-year roadmap from twenty-plus stakeholder interviews. The most fun was AfterWard: an AI recovery companion for surgery patients, designed around three NHS pathways, which won Most Viable Business at Edinburgh Innovations’ Startup Fast Track.

Now, as a strategy and product consultant at Effectus Research, I am turning board-level governance thinking into working software with founder Brian Mooney. Strategy, it turns out, is also something you can ship.