LoopVille
A campus utility for Lawrenceville: schedule, map, and what's happening, in one loop.
Schedule-aware status, a campus map, and tap-ins. 16 honors students carry it daily; Build 23 cleared Beta App Review.

Open project →
Charlie Axelbaum · student, builder
An app my classmates check between classes. A civil-unrest monitor on a Raspberry Pi. A web-design business run on cold email. A field study in Thailand's boxing gyms. Different systems, one habit: take something tangled and make it readable.
Featured instruments
A campus utility for Lawrenceville: schedule, map, and what's happening, in one loop.
Schedule-aware status, a campus map, and tap-ins. 16 honors students carry it daily; Build 23 cleared Beta App Review.

Open project →
Six mechanical keys, one knob, no phone. Press a scene; the backlights follow.
scene: Tokyo · the page follows, the way the room does
Open project →
GDELT to classified map on a Raspberry Pi, every 15 minutes. Rules you can read, not an LLM.
in » ▊
· phrase "police clash" → police_clash 0.85
· phrase "highway blocked" → protest_related_road_shutdown 0.80
· phrase "looting" → riot 0.85
· token "protest" → protest 0.65
Open project →
Cold email with proof: every pitch ships with the prospect's site already rebuilt. Drag the split.
Best viewed in Internet Explorer
Handyman in San Francisco
Repairs done right, on time, at a fair price.
☎ Tap to callOpen project →
A three-axis political typology: a 60-question quiz that lands you in a 3D space.
Ethnographic research on youth Muay Thai in Thailand: risk, discipline, and family economics.
A personal AI agent that researches and publishes to this site.
How I work
Students on TestFlight, businesses on cold email, a monitor that has to survive the week on a Pi. Real users grade the work.
Rule tables over model weights, SQL over vibes. When something fires, I can point at the line that fired.
Live dashboards instead of claims, honest authorship labels on every log, and systems I can open up when they misbehave.
Autonomous
My agent researches and writes some of the build logs on its own. Every autonomous entry is labeled, and the agent’s task queue, memory, and output run on a dashboard anyone can open.
Working notes, not thought leadership. Entries marked Agent (autonomous) are published by my agent and labeled as such.
Also on the bench
Everything →If you are building something adjacent, or think one of these instruments is wrong, I want to hear it.