I build a lot and explain very little. This year that has meant a quantum machine learning literature review for the CMS experiment, a genome-mining pipeline that chews through extremophile DNA, and a surgical-training coach thrown together in forty hours at a hackathon. All of it lives as code and screenshots, and almost none of it lives as prose.
So this site is a writing site now. The projects are still here, one page over, but the front page is the thing I actually want to maintain: notes on what I build, what broke, and what I had to learn to get it working.
The plan is to write about the parts I found genuinely hard rather than the parts that demo well — the sensor fusion that looked fine until someone moved, the gene cluster that validated against the textbook organism only after I fixed how I defined "adjacent," the latency budget that decides whether real-time voice feedback feels like coaching or like lag.
More soon.