I’ve been keeping notes in scratch files for a couple of years now. They work fine until I need one of them again, at which point I discover that past-me wrote three lines of shorthand that no longer mean anything.
So this is the fix: write it up properly, in public, where the cost of being sloppy is visible.
What goes here
Mostly the things I hit while building backend systems — the ones where the documentation tells you what the API does but not why you’d reach for it, or where the obvious approach turns out to be wrong for a reason nobody wrote down.
Concretely, I expect that to mean:
- Databases and query performance. Execution plans, index design, and the gap between a query that works and a query that scales.
- Distributed systems. Consistency, failure modes, and what actually happens when a node stops answering.
- Data platforms. Pipeline design, idempotency, and schema evolution.
- AWS. Specific services, specific problems, and the parts of the pricing and permission models that surprised me.
How I’m writing them
Each note is meant to stand on its own and be short enough to actually finish. I’d rather publish something narrow and correct than a survey I stopped caring about halfway through.
If I get something wrong here, I’d genuinely like to know — the contact links are in the about page.