← Home
Writing
Engineering and research notes from building local-first tools, agent infrastructure, and evaluated ML experiments. The writing focuses on problems, evidence, and lessons rather than private operational recipes.
What Blackreach taught me about reliable browser agents
Reduce noise, preserve progress, verify outcomes, and keep uncertain results visible.
What shared memory taught me about agent coordination
Provenance, relevance, ownership, correction, and deliberate handoffs.
We spent months measuring whether our AI was getting better or worse
Most teams ship agents without knowing if they actually work. I built a tool to fix that.
Linear A, and why I want to take a crack at it
70 years of failed phonetic matching. Why the tools available now are different.
Why I gave my coding assistant a local voice
A narrow local tool worked because it solved one missing part of an existing workflow.
I built a GitHub dashboard for my terminal
BubbleTea, Go, and why I never want to open a browser just to check my notifications again.
Fine-tuning a protein model to find new antibiotics
A leakage-aware protein-language-model experiment, documented results, and a failed approach worth keeping.
Honest results from a small GRPO lab
What held up, what failed to transfer, and why negative results belong in the public record.