How it started
My first hands-on projects were Telegram bots and agent automation. I was setting things up on a remote server, connecting tools, and learning by getting them to run.
Hackathons gave the work a more concrete shape. TuringVault explored inspectable AI decisions on Mantle. OneLink Pay explored what software should be allowed to spend, and how a person could keep control of that permission.
What I’m spending time on
Synrail and RHOOK are ongoing work. With Synrail, the question is what evidence a coding agent needs before calling a task complete. With RHOOK, it is what changes downstream when one historical input is different.
Both involve coming back to the details: verification behavior, examples that someone else can run, and documentation that describes the limits. The updates page links to some of those changes.
Working with agents
AI agents are part of how I build. That experience also shapes the problems I choose to work on. The source code and project pages are here so the work can be inspected, tried, and discussed.
A few milestones
OneLink Pay won the Magic Labs Bonus Challenge at UXmaxx 2026. TuringVault received a Project Deployment Award at the Mantle Turing Test Hackathon and appeared in its AI Trading & Strategy Demo Day.
The project pages have the demos, supporting material, and a little more about the implementation.