hi I’m Prathamesh Patil

the role was unusual. not design - something closer to product and prompt engineering. my job was to refine the prompts inside their AI agent harness to improve output quality. I didn't open Figma once. worked directly in the backend. got to understand how prompts are structured, skills, referencing, LangSmith, LangFuse. made small, big changes to the prompts - raised four PRs in two weeks, two of which got merged. built a frontend clone of the product to test some feature ideas I had.

three weeks in, we wrapped it up early. a few things converged at once - I wasn't fully focused during that period, the role needed deeper technical ownership and faster shipping than I could give. got tons of good feedback.

proud of what I did ship. clear on what I'd do differently. actively filling the gap.

FlytBase builds software for autonomous drone operations - the kind of product where a wrong design decision means a drone going somewhere it shouldn't. no design references exist for this space. no competitors to learn from. everything from first principles.

I joined not knowing the domain at all. had tough times understanding the domain and the intense work environment. never had expected a B2B company to be this fast. within months I was in the leadership team. within a year I was hiring. the place trusts people early and completely - which is either terrifying or the best possible education depending on how you respond to it. I responded well, mostly.

led design end to end - features, design system, the whole product, with a very lean team. then AI hit and we were one of the first teams genuinely rethinking how drone workflows change when machines can reason. ran experiments daily. hosted events. figured things out in public.

FlytBase also gave me a month in San Francisco and tons of opportunities - a dream I didn't know I'd get to have. I'll always be grateful for the time I had here.

left after nearly three years wanting a different domain and a different set of problems. not because anything was wrong. because there's only one life.

small agency in Mumbai. two designers who were exceptionally good at UI, Figma execution and speed. I was bad at both when I arrived - fairly new to the domain entirely. I absorbed what I could in three months. left with sharper Figma skills and a better eye for UI quality.

I wanted a product company, not an agency. the business and product thinking wasn't there. I took the call and moved on. no regrets - it was the right training ground at the right time.

got to work on wide range of hardware products at different stages - including two that were close to shipping. learned the full lifecycle of a hardware product: how parts are sourced, manufactured, shipped, costed. the real lesson was about where the money is made in industrial design in India - and it's not design. it's manufacturing. almost everything sourced from China, design as a thin layer of aesthetics over it.

that was the moment I decided to switch to software. hardware is hard. if I ever want to build something myself, software is the more honest path. grateful to Umang for the exposure at a young age, in a small environment, I got to see how business actually works.