This is the first post on the rebuilt site, so a quick orientation.
What I’ll write about
The intersection of computer architecture, ML systems, and the performance-engineering work I do day to day. Concretely, expect write-ups on:
- Performance engineering pragmatics — workload characterization, hardware counters, perf modeling, the gap between what a microarch spec says and what shipping silicon actually does. Generic enough to be useful; specific enough to be honest.
- ML systems on constrained hardware — energy-harvesting, intermittent computing, computational storage, sustainable edge servers. The threads from my Ph.D. that I’m still curious about.
- Side projects — small SoC perf tools and visualizations I’m building in my spare time.
- Notes from papers I’m reading — short, opinionated. Less “review” more “what I took away and where I disagreed.”
What I won’t write about
- Internal Arm specifics, roadmaps, or anything that wouldn’t pass public-disclosure review. If you’re a fellow engineer interested in talking shop within those bounds, say hi.
- Hot takes on industry drama. There are better venues for that.
- Things I haven’t actually done. I’d rather post less than post bullshit.
Cadence
Roughly monthly, sometimes more often when I’m in the middle of a project. The RSS feed catches both blog posts and news entries, so if you’d rather not poll, subscribe.
That’s the framing. First real post coming soon.