Kuba Martin
Director of R&D Engineering @ Spacelift; various open-source stuff on GH
developer/devops/data tooling, distributed systems, databases, and lots of other tech stuff
Reddit/HN/GH:cube2222
Opinions my own.
- Weekend thoughts on Gas Town, Beads, slop AI browsers, and AI-generated PRs flooding overwhelmed maintainers. I don't think we're ready for our new powers we're wielding. lucumr.pocoo.org/2026/1/18/ag...
- Lecture B02 of Statistical Rethinking 2026. Multilevel model expansion: workflow, cluster and feature engineering, non-centered parameterization, dank memes. Continues next week with models of population covariance and (if there is time) group-level confounding and "Mundlak machines".
- Getting my new slides in shape. I have it on good authority that the typeface I have chosen is "woke" (aka sans serif)
- I'm teaching Statistical Rethinking again starting Jan 2026. This time with live lectures, divided into Beginner and Experienced sections. Will be a lot more work for me, but I hope much better for students. I will record lectures & all will be found at this link: github.com/rmcelreath/s...
- Shielding Chart xkcd.com/3158/
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- New benchmark just dropped: SnitchBench by Theo Browne tests if LLMs will snitch on you to the authorities if you feed them incriminating documents and a tool that lets them send email, as seen in the Claude 4 System Card Turns out they pretty much all will! simonwillison.net/2025/May/31/...
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- I'll get straight to the point. We trained 2 new models. Like BERT, but modern. ModernBERT. Not some hypey GenAI thing, but a proper workhorse model, for retrieval, classification, etc. Real practical stuff. It's much faster, more accurate, longer context, and more useful. 🧵
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
- [Not loaded yet]
