Michael Chen

AI engineer in
the Bay Area.

I build agent runtimes and evaluation harnesses at Epoch AI / Zerg AI, and I do forward-deployed engineering with our customers.

Selected work

Agent systems, speech, and listening.

Epoch AI / Zerg AI · 2025 to now

Zerg is a spec-driven framework for autonomous coding agents, and ZTC is its terminal client.

My work spans parallel tool dispatch, isolated batch execution, and the live web interface.

Speech research · Apple Silicon

I distilled Kokoro-82M into a single-voice audiobook engine for MLX with about 10M active parameters.

Fast and intelligible, but less clear than Kokoro. Both quality gates fail.

Personal software · Audiobooks

I built Epub Listener, Hark and Kestrel to turn books into audio and listen on my device.

On-device narration, read-along, and offline playback. Visual: existing iPhone test capture.

2025 to now
AI engineer, Epoch AI / Zerg AI Agent runtimes, CI on Fly.io, a multi-agent simulation engine and evaluation harnesses, plus forward-deployed work with customers.
2023 to now
Software engineer (part-time), MOBIVOLT AI-agent circuit design in KiCad and ngspice, test-stand software in C# and Avalonia, hardware-in-the-loop simulation and firmware migration for a DOD-funded fuel-cell program.
2023 to 2025
Founder and president, AI at UCR UC Riverside’s first official AI organization. 50+ members and 10+ technical workshops.
June 2026
B.S. Computer Science, UC Riverside GPA 3.66, with a minor in Strategy & Entrepreneurship.

AMD University Program Award

AMD Pervasive AI Developer Contest, for PHiLIP, a human-in-the-loop image-generation prototype I worked on with a team, built on AMD ROCm.

Official results
Cutie Hack 2022

First place in the sustainability track.

Week 5: Three Tight Ends, Two Early Exits
Fantasy Sports · 2026-10-15
Week 4 was a 195.18-point win. In Week 5, I kept most of the same starters and lost 133.34–170.60, dropping to 3–2.
How much orchestration do I actually need?
Tech & Building · 2026-10-09
I just think the useful interface and the size of the system are separate questions.
Teaching a local speech model when to ask for help
Tech & Building · 2026-10-05
A local speech model can ask a stronger model for help and speak its reply while text is still arriving. This is what worked, and where it still falls short.

Away from work, I build apps for friends and personal tools, and I write.
Projects & Fun Writing About me