I use AI tools as part of my workflow and I won't pretend otherwise. What I actually care about is whether I understand what I've shipped.
I tend to build things to actually work through an idea, not just to have something to show for it. If I'm curious about how something works, the fastest way for me to really get it is to build a rough version myself. I'd rather get the boring parts right like error handling, graceful failure, tests that mean something than have a demo that looks good and breaks the moment someone pokes at it. Privacy-first, local-first design isn't just an interest of mine, it's usually a default I build toward.
A lot of my repos exist because a side project was the most honest way to sit with a question I hadn't fully answered yet including, more than once, my own relationship with money.
Click for a random fact about me
- I'll pick apart a bug for hours before admitting I need a break.
- I have more personal-finance trackers on GitHub than actual savings habits.
- I'm more proud of tests passing than features shipping.
- A good chunk of my repos exist because I got mildly annoyed at doing something manually and automated it instead.
What I'm currently curious about
- Agentic AI systems that can actually be trusted to run unattended (compliance, research, orchestration)
- Privacy-first / local-first apps
- Knowledge graphs as a better way to search than keyword matching
Still looking to get better — always happy for my code to get reviewed by people more experienced than me.