Skip to content
MP

About

Mansukh Patidar

I have spent twenty years leading software engineering at scale. I spend my evenings designing hardware. The interesting work is in the gap between those two things.

01The day job

Senior Director of Engineering at Adobe

I lead the Creative Cloud Experience and Engagement group — around 230 engineers responsible for how creative professionals discover Creative Cloud, onboard onto it, and get the full value of it. Twenty-one years at Adobe, from writing code to running organisations, through web platforms, desktop applications, conversational AI, and now agentic AI. Ten US patents and a Distinguished Inventor title along the way.

02Nights and weekends

Hardware engineer, building with AI

I design and build real hardware — schematics, boards, firmware, enclosures, the lot. Right now that means power measurement instruments; the domain shifts with whatever I need and cannot buy. The constant is doing it with AI in the loop at every stage, which changes what one person on a home bench can finish.

03The thing I actually care about

Closing the gap between hardware and software

Software engineering has had thirty years of tooling built for it, and AI slotted straight into that. Hardware has almost none of it. I build the instrumentation and tooling that plugs AI into the hardware design lifecycle — so a board bring-up can run the same tight, measured, closed loop that a codebase already does.

Why the third one matters

A software team ships, measures, and corrects in minutes. A hardware team waits weeks for a board, then debugs it with instruments that cannot tell the AI anything, because nothing they measure is legible to it. That asymmetry is the whole problem. Closing it means building the measurement layer first — instruments whose output an agent can actually reason about — and that is what PowerBench AI is.

What I write about here

  • Personal software — tools built for an audience of one
  • Bench notes, and the wrong conclusions I drew before the measurement
  • Hardware and firmware bring-up, honestly, including the retractions
  • What AI is and is not good for when the thing under test is physical

A bias worth stating

Most of what is here is measurement-first. If a claim on this site is not backed by a number I took myself, I try to say so — and when a measurement later kills something I claimed, I would rather publish the correction than quietly edit the post.

Elsewhere