About me
Hi, I'm Omkar Khadamkar.
I'm a product designer in Pune, India. I design software people can trust, and write the rules AI follows to build it.
- 11 years in design
- Enterprise security since 2020
- Indian, UK and US East hours
- Available now
What I do
Since 2020 I've designed enterprise security software, first at VMware, then at Skyhigh Security. These are the screens where people write security rules, check what the software found, and act when something goes wrong.
Two things are changing: AI builds the screens, and AI agents take actions. Both are good, as long as risk is never guessed and people keep the calls that matter. Designing those moments is my job.
At VMware I designed one screen that shows which of 32 network locations has a problem. It shipped, and I'm the first-named inventor on a published application, US 2025/0036264.
Same job, new readers
- 2013Future Group, fashion retail Read by eleven garment and accessory vendors. None of them worked for me, so the spec had to settle each argument.
- 2020VMware Read by designers and engineers on NSX. The UX guidelines they built against.
- 2022Skyhigh Security Read by designers, engineers and AI tools. The design system I led, rewritten so AI tools build screens that match it.
- NowVela Design System, open source Read by design files, code and AI tools. One set of rules they all follow.
How I work
- Never let AI guess the risk.AI can draft layouts, charts and summaries. The words and colours that say how serious something is stay fixed, and the same on every screen.
- People keep the calls that matter.An AI agent can earn the right to act alone, one action at a time. Deciding what still reaches a person, and when, is the design work.
- Measure first, and show the misses.I measure before I change anything. When a result goes against me, I publish that too.
- Make the right way the easy way.I've never had a direct report. I've coached three designers, none reporting to me. So instead of deciding things, I make the right way easier than the wrong way. Nobody had to do what I said
What I'm exploring
- Letting an agent act aloneWhen should a security lead let an AI agent act without asking first?
- Who starts the EU's 24-hour clock?A new EU law gives software makers 24 hours to warn the authorities about a flaw attackers are using. Who decides when that clock starts?
- When another AI asks your productWhen the user of your security product is another AI, what does a good answer look like?
- Signal, my newsletterEvery two weeks, for people who build security products.
Work with me
I'm open to a full-time principal role in security product design. I also take fixed-price design pilots with security companies.
If something in your product makes people guess, I'd love to hear about it.