Principles
How I work
I pick boring technology when it wins. JustCar needed search, so I tried plain Postgres pattern matching before reaching for a search engine. It handled 49,000 records fine, so Elasticsearch never got added.
I ask what the worst outcome is before I start. That is why I can move fast. The AWS talk I gave in Cebu was mostly about this: build the MVP, bound the bill at zero.
And I write down what I got wrong. 13 of these case studies have a section on what I would do differently.
Ten principles
Remove friction before adding tools.
Design environments and habits instead of writing wish lists.
"I am someone who..." beats "I should...".
Small daily inputs compound, and most of the payoff arrives late.
Track the lead indicators, not just the outcomes.
Start simple.
Gall's Law: the complex systems that work grew out of simple ones that worked.
Write it down.
ADRs, RFCs and docs force clarity and let knowledge outlive you.
AI drafts, I decide.
Read every line and own what ships.
Distribution beats product.
A worse product with better distribution wins, so validate demand before building.
Failure is tuition.
Every failed swing should leave behind something reusable: a validated ICP, a channel, an audience.
These come from my peak-self repo on GitHub.