
Safari King Africa
How I rebuilt a Tanzania-based operator's web presence as a marketing site and a real back-office, with Claude wired into the admin screens where the operator actually works.

I'm an applied AI and data engineer in Seattle.
I build AI features into products, and the data platforms underneath them. Right now that's school software for Tanzanian institutions, a safari operator's booking and back office, and a college basketball analytics platform. Away from the desk I'm usually on a bike or out on a long run.
I’m building school software for Tanzanian institutions, from primary schools through postgraduate. Every school grades differently, fees are structured differently at every level, and the government wants its reports in a particular shape. The AI in it aims practice at what a class actually got wrong, and pseudonymises pupils before a request leaves the country.
My college basketball project. I rebuilt it on a much simpler model after working out that the old one had been leaking for a season, and I’m getting it ready for November.

How I rebuilt a Tanzania-based operator's web presence as a marketing site and a real back-office, with Claude wired into the admin screens where the operator actually works.
A case study from one summer building an AI system inside a Microsoft-native enterprise.
The accuracy figure on my March Madness platform was leaking. What it took to notice, months after I had published it.
A hearing-aware audio platform that measures how a given pair of earplugs changes what you hear, then filters your music to match, built on a Redis-free Postgres job queue.
A medallion pipeline on AWS (Bronze, Silver, Gold) that cut query latency about 16.5× (10.9s to 0.66s) and data scanned from roughly 420 MB to 2 KB, built as a graduate course capstone (DATA 516) on synthetic data.
I watch an unreasonable amount of it. Seahawks, Mariners, college basketball, Manchester United, and most of whatever else happens to be on.


Running, biking, and long walks. Mostly this is how I get away from a screen for a couple of hours.
There is usually something I am partway through, and a longer list of things people keep telling me to watch.
I’m always happy to talk about technology, ideas, or whatever you’re building.
I also work with teams now and then on AI and software problems. If you’re building something and think I could help, say hello.
Essays and tutorials on AI, data, and building things.