Writing · 3 posts
Thinking out loud.
On building, shipping, and what it means to run an AI-native boutique studio with agentic coding.
The Software House of One
How a single engineer can ship like a five-person team with Claude Code and a spec discipline.
AgencySDD
May 28, 202611mEvaluating the eval
I added tool-calling to my chatbot, broke the test suite without realizing, and almost shipped it — a small story about the discipline LLM apps demand.
EvalsAI
May 25, 20266mLLM-as-Judge: A Tasting Notes Approach
How I built an editorial QA pipeline that doesn't gaslight me.
PipelinesLLMEditorial
May 25, 202610m