Pradeep Kaushik

At NVIDIA, I work in product and developer relations across Physical AI and Agentic Intelligence. I care about the messy part: getting AI to take dependable action in the real world. That pulls me into robotics, simulation, digital twins, multi-agent systems, and platforms like Omniverse. It's where a model stops generating text and starts moving something.

I started on plant floors, commissioning automation systems. From there I moved into the MES and industrial IoT software around those systems, and the analytics on top. Most of that work happened over a long stretch at Rockwell Automation.

At AWS, my work moved to cloud scale: digital twins, analytics and simulation, then generative and agentic AI. For more than twenty-two years, I've watched the tools change while many of the industrial problems stayed the same. That is why I care less about what a model can demo and more about whether it holds up on a factory floor at 2am. I spend most of my time in the gap between what gets announced and what developers can actually put to work.

This site is where I write things out in the open. Not a portfolio, not a newsletter. The name is literal: rough cuts from what I'm figuring out about robot learning, simulation, digital twins, developer tools, and what it takes to turn a good demo into a dependable system. I add notes when I have something worth working through, not on a schedule. I also keep a small list of things I think are worth reading.

If something here is useful to you, that's the whole point.

This is a personal site, so to be clear: the views and opinions expressed here are my own and do not represent those of my employer, NVIDIA.

Based in San Jose, California.

---