Our conviction

The next generation of AI should learn from human experience.

Robots that can work alongside people could change how much we’re able to make, repair, grow, and build. We believe reaching that future depends on learning the dexterity, judgment, and adaptability that skilled people develop through years of practice.

That knowledge lives in factories, farms, and workshops around the world, but much of it has never been captured for machines to learn from. We’re building Outpost to turn skilled human experience into the data foundation for physical AI, preserving the actions, context, and variation that make real work complex.

Who we are

Vikram Battalapalli

Cofounder

Vikram grew up building robots, starting in seventh grade as one of the youngest members of his FIRST robotics team.

Then, he went on to build deep learning models for the Navy at the Johns Hopkins Applied Physics Laboratory while still in high school. He continued exploring how intelligent systems act in the world by building self-driving control systems at the University of Illinois Urbana-Champaign, and graduated with a degree in computer engineering in 2025.

Now he’s building Outpost to give the next generation of robots the human experience they need to learn, drawing on his years of work in robotics and machine learning.

Hemanth Bellala

Cofounder

Hemanth is a founder and engineer who’s built companies and the systems that power them. At Georgia Tech, he focused on deep learning and RL while studying computer science. He dropped out of his master’s program to build a home food marketplace startup which he bootstrapped to $240K ARR.

He carried that experience of building from scratch into his role as Head of Data at FIXD, where he built the company’s full customer data platform. At J.P. Morgan, he worked on its new tokenization platform as a senior software engineer.

Now he’s bringing that experience back to physical AI, building Outpost to help robots learn from the people who know their work best.

We’re drawn to hard problems and work naturally together. Now we’re combining our experience in robotics, machine learning, engineering, and building companies to build the data platform for physical AI.

Backed by