I like having an idea, building a version of it, and seeing where it goes.

Most of my projects start with a question or something I want to improve. I make a first version, use it, and keep changing it as I learn. Some stay small. Others become tools I share and maintain.

Right now, I'm exploring:

• MCP and infrastructure for AI agents
• context efficiency, reliability, and security
• search and research workflows for agents
• evaluating AI tools with reproducible evidence

I also contribute to open source and write about what I build, what failed, and what I learned.

→ Personal site → Personal email