Current focus
Reducing alert noise without losing the signal.
Three weeks into pruning a runbook nobody read. The interesting question is which alerts deserve a name, not which ones we can mute.
Engineering log · 2026·04
A patient record of how I actually build things.
I'm a software engineer who'd rather show the trail than the trophy. This site is a small, dated log of work, decisions, and the lessons that earned their place. Read it like a notebook — slowly, and out of order if you like.
// now
Where my attention is.
Quietly learning
Functional reactive ideas in non-FRP languages.
Reading Naur and Hancock back-to-back. Not because I'll write any of it — because the framing helps when the team's mental model drifts.
Bench
A small CLI for keeping decisions traceable.
`decide` writes a one-page ADR-style note, links it to the change, and leaves a breadcrumb that survives the team. Mostly for me. So far.
// log
Recent entries.
Plain notes. Unfinished welcome.
-
On second-best decisions
The decision that wins is rarely the optimal one. It's the one a team can keep agreeing with through three reorganizations. A note on why I've stopped trying to win arguments at design review.
-
Three signals are usually enough
Saturation, latency, and error budget — done well — beat a dashboard wall. A short note on what we removed and what survived.
-
The week the wrong abstraction taught me everything
Spent a week with a generic adapter that turned out to be a junk drawer. Tore it out. Wrote down what I should have asked first.
-
Reading A Pattern Language again, slowly
Notes from chapter 5. Why reaching for the right "alcove" in software feels exactly like a room you can finally sit in.
// decisions
Decisions worth remembering.
A small archive of thinking I'd like to keep honest.
- DEC·011 held
Don't optimize for the resume
Choosing tools the next maintainer can read on a quiet afternoon, not the ones that look ambitious in a hiring funnel.
- DEC·010 held
Write the README before the code
If the README is hard, the code will be harder. Drafting prose first finds the seams the diagram hides.
- DEC·009 retired
Cap engineering reviews at thirty minutes
Long meetings hide weak proposals. Small windows make the strong ones obvious. We get shorter docs and better arguments.
// channels
Reasonable ways to reach me.
- Email hello@stergioulas.com
Best for slower conversations and longer threads.
- LinkedIn linkedin.com/in/petrostergioulas
Roles, references, and the timeline this site won't keep.
- RSS /feed.xml (soon)
For when this becomes a small, reasonable feed.