Perspective · 2 min read
Where I’m from
And what it changes about the work.
The Central Valley
In rural places like the Central Valley, a lot of people live a long drive from whatever a product assumes is close by: the bank, the club, the office that takes certified funds. I keep them in mind. They keep turning up in the work: renters driving 15–30 minutes for certified funds, and players good enough for college soccer who grow up too far from the big clubs for a recruiter to see them.
The Choctaw Nation of Oklahoma
Proud future ancestor of the Choctaw Nation of Oklahoma. I try to work like one: leave the people after me something they can read, trace and keep running without me. Written specs, numbers with their sources, systems the next person can follow.
I left a job over the data
At Oracle the company was pulling together location data, browsing behavior and real-world signals to predict what people would buy. I had growing concerns about where that data came from and how it was being used, and they eventually led me to leave. If the data you have isn’t data you should use, I’ll say so.
A recovering data scientist
I still count everything, then I poke at the count. At AppFolio, 44 of every 100 started rental applications were submitted. Every applicant was anonymous, so the number couldn’t say who stopped, or where.
I build my own things
An apparel company, back when. Brava, with my brother. Six free tools on this site that someone’s AI agent can run. I’d rather show you a working thing than a deck.
The unglamorous middle
The customer calls, the half-clean data, the code that has to actually run. That’s where I’m useful, long before the demo.
A few at a time
I take on a few clients at a time, on purpose. Each one gets real attention, and I stay close enough to the work to do it myself.
Selected work
- AppFolio, applications
- Rebuilt the rental application end to end. Submissions went from 44% to 70%. The story
- AppFolio, payments
- Checkout for application fees, with a recovery path for failed payments. 99.1% of application payments completed. Résumé
- Lively
- Built the money-movement platform, and moved $50M+ in HSA assets between financial institutions with zero data loss. Résumé
- Oracle
- Built a system to run hundreds of evaluations against model updates, so a change was judged on real statistical significance, not gut feelings. The story
- Brava
- With my brother: the AI production pipeline, and a COPPA and privacy review for minors’ data. bravamade.com
- AbiertoClaw
- Finds the best free AI model each day with a real tool-call test. On GitHub
- Native Plant Finder
- A free tool that someone’s AI agent can run, by link or over WebMCP. Open it
- This site
- Plain HTML, CSS and JavaScript, with no framework. The source
The first conversation is free, and it’s the quickest way to find out whether we’d work well together.