I’m at the U.S. Census Bureau with U.S. Tech Force, working on the app that keeps the Bureau’s address and map database up to date. Before this, three years at JPMorgan Chase, and before that, UPS.
Work
U.S. Census Bureau
IT Specialist (GS-14), U.S. Tech Force fellow
I inherited a proof of concept for the MAF/TIGER Update System, the replacement for the Geography Division’s old update application, and turned it into an MVP. I finished Phase 1 in three weeks with AI, a month before it was due, and I’m now on Phase 2.
- Connected it to the Bureau’s core API and restructured the code around domain-driven design
- Got access control down from 30 roles to 4 without losing any permissions the app needs
- Brought over the agentic development lifecycle I built for my own projects with Claude. Every change goes through the same steps:
- Spec
- Issues, by priority
- Implementation
- AI self-review
- CI
- Three agent reviews
- Auto-merge
- Product owner sign-off
JPMorgan Chase
Software Engineer II, previously Software Engineer I
Lead developer on a rule engine that checks thousands of production ERwin data models. A few hundred people use it each week.
- Reports that took 10–15 minutes in the old app now take under 10 seconds
- Went from 84 rules to 65 while adding 10 new ones, and fixed false positives in 5 of them
- Before that, worked on the firm’s data catalog (nested data structures, a Kafka integration with the data lake) and was promoted early
UPS
Engineering Specialist, then Engineering Supervisor
Ran maintenance teams of up to 15 people at a sorting hub that moves 65,000 packages an hour, and managed a few six-figure capital projects.
University at Buffalo
BS, Computer Engineering
Side projects
Most of what I’ve built since September 2026, using the agentic development lifecycle I designed and then brought to the Census Bureau. None of it is public yet, so the names stay blacked out.
Voter information
LivePut in your address and it shows every race on your ballot and where to vote, with an archive of election data behind it.
Next.js, Python, Postgres
Transit metrics
LivePulls a transit agency’s real-time feeds every few seconds and works out on-time performance, headways and a reliability score for each line. Also draws every train and bus on a live map.
Python, FastAPI, MapLibre
Grocery prices
LiveTracks prices at the stores near me and tells me when something I buy goes on sale. AI agents match the same product across stores from its listing data and photos.
FastAPI, Postgres, Claude API, a Chrome extension
Newsroom archive
LiveA web archive for a small group of journalists. It saves pages and video and keeps proof of when each one was captured.
Django, Postgres
Net worth
A net worth app where the data stays on my own devices instead of going through a bank aggregator.
TypeScript, React Native
iPhone AR
Measuring, placing objects, drawing, and object detection on LiDAR phones. I don’t own a Mac, so it builds and signs entirely in CI.
Swift, ARKit, Core ML
Emulator tuning
Got an open-source Android emulator from around 60 fps to a steady 118.
C, QEMU, Vulkan
Media archiver
A self-hosted archiver that checks sources on a schedule.
Python
Tools
At work it’s mostly Java and Spring Boot with React, TypeScript and Oracle, plus Kafka, Kubernetes and Cloud Foundry at JPMorgan. On my own projects I use a lot of Python, TypeScript and Postgres. I use Claude Code daily.
Older, on GitHub
- 16-bit RISC CPUVHDL
- PINTOS threadsC
- Driver attention monitoringPython, OpenCV
- Ray tracerC++
- Tetris for the terminalPython