Nuclear reactors
to neural
networks.
I have spent a career mastering progressively complex, high-stakes systems: nuclear propulsion, then global payments, now generative AI and agentic systems architecture.
One ecosystem, four bonds.
Peppi Labs is the studio: rapid prototyping, full-stack builds and AI architecture, on a two-week sprint that ends in a ship-or-pass decision. The same team then turned that method on itself. Peppi.ai is a personal AI companion that works entirely over SMS, no app. AskPeppi gives a local business an assistant that answers customers around the clock. PeppiPay is spending infrastructure for AI agents, with real-time policy enforcement and a human in the loop. Consulting, then three products we designed, built and now operate ourselves.
Nine things I built and run.
These are my personal projects, built, funded and operated by me outside client work. Peppi Labs engagements are separate.
Where a rollback wasn't an option.
Nuclear reactor operations and regulated money movement teach the same discipline: you cannot ship optimism. Every assumption has a failure mode, someone has to own it, and the person who owns it writes it down. AI collapses the time between an idea and a working system, but speed without governance is just a faster way to be wrong. The controls carry over. Oversight, auditability, a human in the loop at the decisions that matter.
Startup advisory & mentorship.
Founders I work with directly on product, payments architecture and the discipline of getting to a decision.
It started with materials science.
As an undergraduate at the University of Florida I published peer-reviewed research in advanced materials science, fully funded by Pratt & Whitney to pioneer next-generation aerospace composites. Systems-level thinking about how things fail has been the through-line ever since.
Writing.
Post Agile AI Development
Building at the Speed of Thought
Building at the Speed of Thought Adapting software development processes to match AI’s acceleration There’s a gap every product team knows well — the space between having an idea and s...