A prevention program’s year, start to finish
Plan a class on the PI-DPP schedule, capture its sessions from paper or a spreadsheet, and watch outcomes, CDC Full recognition and Medicare revenue add up.
Lukio is the data system I volunteered my time to build for organizations delivering sustainable health programs for underserved communities in the U.S.-Affiliated Pacific Islands. It makes the programs easier to run, improves them through dashboards and insights, and gives new staff in any role helpful tools sooner.
I'm open to software engineering roles now.
Data systems, internal tools and apps, for teams with sensitive data and busy staff.
In use today by diabetes prevention programs, from a person's registration to the file the program sends CDC.
Staff work from a desktop or a phone and keep working when the connection drops. Participants have their own phone app, and a photo of a paper form fills in the data for staff to check.
Dashboards and insights show each class, site and program how it's doing. The CDC submission file is built and checked before it goes.
Eligibility, session schedules and CDC's recognition standards are built in for each program, so new staff in any role start with helpful tools.
Each program gets a participant app set up from its own plan and goals: the next session, check-ins, messages to the coach and progress.
I joined AAPCHO in 2018 doing data entry for the Pacific Islander Diabetes Prevention Program (PI-DPP), saw what its data needed, and built my role as its data evaluator around that. There was no data system, so I built Benji from scratch in Google Sheets and Apps Script, for 11 organizations across the U.S., Palau, the Federated States of Micronesia, the Marshall Islands and the Northern Mariana Islands.
On site in Chuuk, Kosrae, Ebeye, Palau and Saipan, I mapped each program’s workflow from the session to the report and traced data problems to their root causes.
Lukio starts from the paper session log coaches already keep, and a photo of it fills in the session.
My cleaning process flagged problems for correction, based on its configuration.
Lukio checks each entry as it is made and marks a weight that jumps for a look before it is saved.
One workflow, taught hands-on, in person and online: record the session log on the day, then enter it.
The same workflow is the app’s: plain words, values filled in, one Save.
Every six months I tallied each site’s results against milestones modeled on Medicare’s billing codes. More than $1 million went to sites over six years.
Lukio tallies Medicare billing codes from the sessions themselves.
My Benji Processor showed each program where it stood before CDC’s evaluation.
Lukio shows each class’s CDC recognition standing before the file goes.
Each organization worked in its own secured folders.
Lukio’s database lets each organization see only its own records.
Four working demos with sample data. Each opens straight into the demo.
Plan a class on the PI-DPP schedule, capture its sessions from paper or a spreadsheet, and watch outcomes, CDC Full recognition and Medicare revenue add up.
A tug and barge fleet's supply requests and maintenance, set up on the same platform with no engine changes.
Sign in the way every staff member does: a password, then a code from an authenticator app.
Sign in as a staff member and try reading and saving across organizations. The database allows or refuses each, and says why.
How a program went from paper to 96% CDC-acceptable data in a year: the field visits, the checks and the training.
Data systems that replace paper and spreadsheets, dashboards, internal tools, and HIPAA-aware apps that keep working offline.
I designed and built Lukio, a program data platform for many products, as my own project. Lukio Health is its first product, starting with the CDC National Diabetes Prevention Program. Pasifika Health Data Center, the program data product I built for AAPCHO, ran on it as Penji.
Built Benji, the data system that took the Pacific Islander Diabetes Prevention Program from paper to digital, and raised the share of participant data CDC could accept from about one in five to 96% within a year.
Built a working prototype of a supply request system for a tug and barge fleet.
Front-desk appointments: quick computer fixes, consultations and customer questions.
Led deck maintenance crews for a tug and barge fleet in Honolulu.
I have been teaching myself since I was a kid, from rebuilding web pages in Notepad to running game servers at 13. I build tools for people, and I test them while I build them: I use what I make, notice what gets in the way, and refine it.
I have been teaching myself since I was a kid. I learned HTML by viewing the source of websites and rebuilding what I found in Notepad, then used FrontPage, Dreamweaver and Photoshop to experiment further. At 13 I was running EverQuest emulators and built a working PC out of broken ones.
Helping people with technology started early, too. In the mid-2000s I wrote a guide to hosting your own dedicated Counter-Strike server, and the official Valve forums for Counter-Strike pinned it for other players. I went looking for people with computer problems on forums and helped them for free, aiming to fix each problem without anyone buying new parts. It was all done in writing: about half of them had a microphone, and nobody had a camera. Years later, at the Geek Squad front desk at Best Buy, the hardest part of the job was keeping appointments to their allotted time, because I wanted to answer every question and pass along every tip.
None of it was formal. I learned what the work of the moment or my curiosity called for, mostly from examples I took apart and got running myself. Documentation came late: I could not really read it until about 2020. Since then I have added Bjarne Stroustrup's The C++ Programming Language, the official documentation for the tools I use and thousands of hours of study, and I still study every day.
I enjoy learning how a tool will be used: where it will live, who will use it, what it replaces and whatever else shapes it. A tool has to fit the people who use it and the environment they work in. The platform I developed under the name Penji, now being rolled out as Lukio, is built for coordinators across the U.S.-Affiliated Pacific Islands, where many connections run at 28 kbps and drop in and out. It works online and holds its state through each dropout, so nobody loses work in the middle of a form, and it installs from a download under 5 MB. The aim is a tool that is ergonomic, correct, useful and efficient, and that makes the work it serves easier.
I look for the right solution. I don't always find it on the first attempt. I close in on it quickly: I adjust my sights to your parameters, then do the field work and the study the problem calls for. When I take something on, I go deep, often late into the night, until I understand it in full.
I build the AI tools my work needs, like agent teams, councils and review processes. I also keep an eye on what others are building, and when a tool of real production quality serves the need, I adopt it. I like building my own, using what's out there and finding out what other people are doing. There's a lot I don't know, even though I work to keep up. Angular ships updates regularly, the rest of programming keeps moving, and there is always more community and open-source work and more projects to explore. The learning never stops, and I remain self-driven and self-taught.
Outside of work I am building Sands, a multiplayer game in Unreal Engine 5, from its C++ gameplay systems and peer-to-peer networking to its 3D art. I also build tools and mods for existing software, often with AI assistance. One of them, EQ Observer, is a companion app for players of an online game first released in 1999.