I started as a petroleum engineer at IIT Dhanbad. Built automations, workflows, and surveillance systems for oil fields across four continents — usually by adapting what someone else had already made and filling in what was missing.
ISB, then McKinsey, then product. Not because I stopped wanting to build — because I realised product is where decisions multiply into thousands of outcomes you'll never directly see. That leverage is what I was looking for.
Amazon. Prime Video. Smartsheet. Each time the same pattern: understand the system deeply, find the part that doesn't work the way anyone says it does, fix it at the root. I've shipped features used by tens of millions of people and killed projects that would have wasted a year of engineering time. Both feel like progress.
AI didn't make me a builder.
It removed the ceiling on what I could build alone.
That's a different thing. mightwork.io is what happens when that ceiling goes away — a record of experiments, half-formed ideas tested to conclusion, and the things I learned that don't fit anywhere else. Some worked. Some are still working. The ones that didn't are usually the most interesting.