I build small projects in public. One problem at a time.
I run a simple loop: problem -> focused build -> micro-launch -> result check.
Each project is small, practical, and measured by real usage or saved time.
Focused workflow tools, internal utilities, and category-specific experiments that remove recurring manual work.
No. AI is used when it helps the outcome. The default is simple and reliable first.
Each project has a proof metric. If there is no signal, it is revised or archived.