Michael Chen

AMD Pervasive AI Winner | AI Engineer

Building production AI systems that solve real problems

scroll
Michael Chen

About Me

AI Engineer specializing in production LLM systems, RAG architectures, and autonomous agents. Currently building AI coding agents at Epoch and developed production test systems for DOD-funded programs at MOBIVOLT.

Founder/President of AI at UCR (AIR), where I scaled the organization from 0 to 50+ members. Pursuing B.S. Computer Science (Minor: Entrepreneurship and Strategy) at UC Riverside (GPA: 3.72, Dean's Honor List).

AMD Pervasive AI Winner (8,200+ participants)

Experience

Professional roles and leadership

Oct 2025 - Present
Remote

Artificial Intelligence Engineer

Epoch (Contract)

Collaborated on Zerg AI, an autonomous coding agent framework featuring spec-driven TDD workflows, world model state tracking, incremental context optimization, and multi-environment execution. Built ZTC, a production TUI orchestrating 20+ execution phases with hybrid RAG retrieval, streaming LLM integration, and SWE-bench evaluation workflows.

20+ execution phases Hybrid RAG SWE-bench evaluation
Jun 2023 - Sep 2025
Livermore, CA

Software Engineer

MOBIVOLT LLC

Developed 10+ automated test systems for $5M+ DOD-funded fuel cell program using C#/.NET, integrating 250+ sensors with zero downtime across 5,000+ hours of testing. Accelerated data analysis by 300% through automated Python pipelines and embedded Arduino firmware (RS-232/TTL/SPI at 50Hz).

5,000+ hrs uptime 250+ sensors 300% faster analysis $5M+ DOD program
Oct 2023 - Sep 2025
Riverside, CA

Founder/President

Artificial Intelligence at UCR (AIR)

Established UCR's first official AI organization, scaling membership from 0 to 50+ active members within 6 months. Delivered 10+ technical workshops on AI/ML implementation and emerging technologies for cross-functional teams, establishing UCR's AI community network and collaborative development culture.

0 to 50+ members 10+ workshops 45% attendance growth

Projects

Personal and award-winning work

AMD Winner

PHiLIP

Human-in-loop image generation platform achieving 2-second inference with 71 style presets, Stable Diffusion upscaling, and iterative refinement. Presented live demo at AMD Advancing AI Event 2024, San Francisco.

2-sec inference 71 styles 8,200+ participants
Python PyTorch HuggingFace AMD ROCm
View Project
Production SaaS

Castor

AI-powered GEO consultation platform with agentic RAG system featuring 150,000-word knowledge base from 15+ peer-reviewed papers and 80+ industry sources with hybrid search and Cohere reranking.

>0.85 faithfulness >0.75 precision 150K+ words
Next.js 15 TypeScript Supabase RAG Stripe
View Project

Skills

Technologies I work with

AI & Machine Learning

LLMs RAG AI Agents PyTorch TensorFlow Langchain HuggingFace MCP NLP Vector Embeddings

Full-Stack

Next.js TypeScript React Node.js PostgreSQL Supabase Pinecone GraphQL REST APIs

Infrastructure

AWS GCP Docker Kubernetes CI/CD Vercel Git Linux

Get In Touch

Open to opportunities and collaborations