Skip to content
View mchen04's full-sized avatar

Block or report mchen04

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
mchen04/README.md

Michael Chen

AI software engineer at Epoch building autonomous coding-agent platforms, LLM evaluation harnesses, and multi-agent simulation infrastructure. CS student at UC Riverside, graduating June 2026, with a minor in Entrepreneurship & Strategy.

My work sits at the intersection of agentic developer tooling, AI infrastructure, and production systems engineering. Recently, I have worked on Zerg and ZTC: an autonomous coding-agent platform and Claude Code-style terminal with parallel tool dispatch, background sub-agents, worktree-isolated batch execution, provider routing, and live web-mirror observability.

I also build continual-validation and embedded evaluation harnesses for production AI applications, scoring outputs against deterministic state, tool, artifact, browser, and visual oracles. Previously, I built automated test systems at MOBIVOLT for a USD 5M+ DOD-funded fuel-cell program, integrating 250+ sensors across 5,000+ hours of testing, and founded AI@UCR, UCR's first official AI organization.

Focus

  • Autonomous coding agents and terminal-native developer tooling
  • LLM evaluation, SWE-bench-style harnesses, and behavior regression testing
  • Multi-agent simulation infrastructure for long-running enterprise workflows
  • Hardware-in-the-loop simulation, test automation, and firmware-adjacent systems

Featured work

  • Zerg / ZTC — autonomous coding-agent platform and agentic terminal with parallel tool dispatch, background sub-agents, isolated batch execution, provider routing, and web-mirror observability.
  • Continual validation — embedded AI evaluation infrastructure using deterministic state, tool, artifact, browser, and visual oracles for behavior regression testing.
  • MOBIVOLT test systems — C#/.NET and Python/PyQt automated test and HIL simulation systems for fuel-cell hardware, 250+ sensors, serial firmware bridges, and long-duration data collection.
  • PHiLIPAMD University Program Award-winning human-in-the-loop image-generation platform with 2-second inference, Stable Diffusion upscaling, 71 style presets, and PixArt-alpha fine-tuning on AMD ROCm.

Background

  • Epoch — AI SWE Engineer working on coding agents, terminal tooling, evaluation systems, and multi-agent simulations.
  • MOBIVOLT LLC — Software Engineer Intern building fuel-cell test automation, hardware simulation, and PIC32MZ firmware migration support.
  • AI at UCR — Founder and President of UCR's first official AI organization, scaled to 50+ active members and delivered 10+ technical workshops.
  • UC Riverside — B.S. Computer Science, minor in Entrepreneurship & Strategy, Dean's Honor List, GPA 3.66/4.0.

Awards

Stack

AI/ML: LLMs, AI agents, tool use, RAG, MCP, prompt engineering, LLM evaluation, SWE-bench, multi-agent simulation, LangChain, LangGraph, PyTorch, HuggingFace, TensorFlow, Scikit-Learn, vector embeddings

Languages: Python, TypeScript, JavaScript, C++, C, Java, C#, Go, Rust, SQL, Bash

Backend: Node.js, Express.js, FastAPI, Django, Flask, REST APIs, GraphQL, WebSockets, Server-Sent Events, Socket.IO

Frontend: React, Next.js, Tailwind CSS, shadcn/ui, Vite, server-side rendering

Databases & infra: PostgreSQL, Supabase, Redis, MongoDB, Pinecone, pgvector, Docker, Kubernetes, k3d, AWS, GCP, Azure, Vercel, Modal, Fly.io, GitHub Actions, CI/CD

Links

Portfolio · LinkedIn · GitHub

Pinned Loading

  1. PHiLIP PHiLIP Public

    AI image generation and enhancement pipeline using generative models with configurable style prompts and automated generation loops.

    Python

  2. claude-switch-application claude-switch-application Public

    macOS CLI for sub-second switching between Claude Code accounts via Keychain swap, with optional shared config and per-profile usage dashboard.

    Rust

  3. castor castor Public

    Subscription-based GEO chatbot platform helping businesses optimize content for AI search engines like ChatGPT and Perplexity.

    TypeScript

  4. ROCm-Doctor ROCm-Doctor Public

    ROCm Doctor: self-healing model CI/CD for AMD MI300X. Continuously supervises a vLLM deployment, classifies operator-change vs drift, and heals via 20 audited recipes. AMD Developer Hackathon entry.

    Python

  5. balldontlie balldontlie Public

    Go bot that finds +EV NBA betting opportunities on Kalshi by aggregating sportsbook consensus odds and applying Kelly Criterion sizing.

    Go 1

  6. Dynasty_Model Dynasty_Model Public

    NBA dynasty fantasy projection model using LightGBM, Transformer embeddings, and VORP-based valuation across 10 seasons of walk-forward validation.

    Python