Systems builderAI developerOperator

I build systems
that think.

From embedded boards to edge AI, I make ambitious technology useful in the real world—where hardware, intelligence, and human instinct meet.

Explore the work GitHub
Agent systems✦Embedded Linux✦Edge intelligence✦Security research✦Operations✦

Selected systems / 2026

Work that leaves the lab.

Software, hardware, and operating systems built to perform under real constraints—not just look good in a demo.

01Field system

PlumeCast

Operational intelligence for hazardous plumes—turning field inputs into a clearer picture of what is moving, where, and who needs to know.

  • Geospatial
  • Decision support
  • TypeScript
Forecasting under pressure
02Field report

Kimi K3 / One Lane

A 2.78-trillion-parameter model on a consumer desktop—and the single PCIe lane that cost 3×. Real measurements, bottlenecks, and lessons.

  • Inference
  • Hardware
  • Performance
2.78T params 1 PCIe lane
03Open source

AI Desktop / Linux

Making modern AI tools feel native on Ubuntu and Wayland—from packaged desktop apps to local-model workflows that people can actually install.

  • Ubuntu
  • Wayland
  • Local AI
Useful beats “supported”
04Systems work

AOSP & Edge Devices

Custom firmware, BSP work, and system-level modification across Allwinner and Rockchip platforms—where software meets silicon.

  • AOSP
  • ARM / SoC
  • Firmware
Board to boot end to end
Browse all 85 public repositories

Operating system / Mine

Technology should adapt to people—not the other way around.

I’m a systems builder, AI explorer, and operations leader. My work moves between terminals, embedded boards, model runtimes, and the real-world environments where the cost of a bad interface shows up immediately.

OmniROM team memberEarly AI beta testerTop-100 operations leaderDetroit area
A

Think in systems

I look for the hidden dependency, the human constraint, and the failure mode before they become the problem.

B

Operate in reality

Years of frontline operations leadership keep every technical decision grounded in what people actually need.

C

Ship the learning

Prototype, measure, document, improve. Curiosity matters most when it produces something another person can use.

Have a real technical challenge?Signal open / 2026

Bring me the
hard problem.

I’m interested in systems work, applied AI, embedded platforms, safety-critical tools, and ideas with enough ambition to be useful.

Start a conversation