Founder Journey
My work is making sure humans stay in charge as our systems get smarter.
AI is moving off the screen and into surgical robots, factory cobots, and brain-computer interfaces- machines that act in the physical world, next to people. These systems keep learning and changing after they ship. But safety approval still happens once, on paper, before launch.
I founded Barbascumpa to close that gap. We’re building the live safety guardrail and flight recorder that lets robotics teams ship AI updates fast while proving a human is still in command.
How We Can Work Together
Become a Design Partner
For engineering and autonomy teams at surgical robotics, collaborative robotics, and neurotech companies who ship AI or control updates and feel the drag of proving each one is still safe.
What you get: Early access to what we’re building, a direct line to shape the roadmap around your system, and a working session each month on your toughest handoffs between people and machines.
Join the Builders’ Workshop (Spring 2027)
A small, invite-only workshop for robotics engineering and QA/RA leaders. It’s four live sessions, and everything shared stays unattributed (Chatham House rule).
What you leave with: A map of every point where command passes between your team and your machine, your release pipeline tested against how regulators expect AI updates to be controlled, and a one-page human authority blueprint for your system.
Speaking & Teaching
For conferences, universities, boards, and policy forums exploring a hard question: when machines act, who answers?
Topics include: liability and the myth of machine personhood, human authority in autonomous systems, and keeping humans in command of surgical robots and neural interfaces.
Where This Work Began
NYU M.S. Capstone: Trust & Safety for AI-Driven Surgical Robots
New York University | May 2025 – May 2026
AI-driven surgical robots often stay in hospitals for more than a decade, connected to hospital networks, health records, and clinical workflows that keep changing around them. Manufacturers design them under one set of rules; hospitals operate them under another. In that gap, responsibility fragments: patches get delayed, incident response is unclear, and no one can easily show who was in command when something went wrong.
Using the NIST AI Risk Management Framework, my research developed a lifecycle playbook for AI-driven surgical robots. It covers risk tiering, cybersecurity, privacy, documentation, and responsibility across U.S. and EU rules.
The research surfaced the question Barbascumpa is built to answer: once a robot is in the field and its software keeps changing, how do you prove, at every moment, that a human is still in command?
M.S., NYU · LL.M., Cardozo School of Law · Harvard Business School Foundry · GCRAI Global AI Ambassador
Code can execute processes. Only humans can answer for them.