Insights
The reasoning behind how I choose problems, design for leverage, preserve human judgment, and evaluate whether a system is useful.
AI-assisted engineering • 5 minute read •
Five failures a passing test suite will not catch — and why four of them hold on their own while one quietly rots.
Read the noteAI-assisted engineering • 5 minute read •
The AI model is the river and your tests are the banks — you can't make the river predictable, only bounded.
Read the noteTechnical article • 5 minute read •
I still copy context between my AI agents by hand. A durable governance pattern already fixes most of it — I run it on PingStep today — and DogBuild is the local tool closing the rest: the actual handoff.
Read the articleEngineering philosophy • 4 minute read •
Four durable principles for reducing friction without turning every problem into more software.
Read the philosophyEngineering note • 5 minute read •
A practical test for recognizing repeated work that should become a system, a delegated responsibility, or an automated workflow.
Read the noteDistributed systems • 4 minute read •
A decision heuristic for tracking the failure modes and responsibilities that an optimization moves elsewhere.
Read the noteDistributed systems • 5 minute read •
A limited analogy for understanding blockchain as a shared-history and consensus problem.
Read the articleEngineering leadership • 4 minute read •
A practical standard for making meaningful issues, risk, and next actions visible before someone has to ask.
Read the noteOperational intelligence • 5 minute read •
How comparable release snapshots can reveal longer-term patterns without replacing engineering judgment.
Read the note