I am learning most of this as I go — data, systems, attention, a few places I got wrong. When something finally makes sense, I write it down here.
Working out when splitting a system into services stops paying for itself — and how you would prove it. My MSc dissertation, currently mid-measurement.
Ugly, undocumented, genuinely useful. The write-up is mostly about why I stopped adding features to it.
Two weeks in, the idea did not survive its first real user. It stays here because the post-mortem was the useful part.
Three years of tuning notes and reviews taught me less than two months of publishing things I wasn't ready to publish.
An explanation I wish someone had given me before I read three papers about it.
What actually got faster, what I broke, and the four commands that did most of the work.
Engineer by training, curious by default. Aerospace, then an MBA in business analytics, now an MSc and a dissertation I talk about too much.
More about meLeave your email and I'll send the next piece when it's ready. That's the whole arrangement — no schedule to keep up with.