Electronic & Computer Engineering · ZJU-UIUC Institute
I'm Edison Liu. I design and ship AI agents, automation pipelines, and web apps, and I understand the layers underneath them — circuits, logic gates, and embedded systems. I don't just patch the surface.
I'm an Electronic and Computer Engineering student at the ZJU-UIUC Institute — a joint programme between Zhejiang University and the University of Illinois Urbana-Champaign.
My background is unusual for someone who builds software: I came up through hardware. I've built a motorised car from a bare breadboard, designed a working vending machine out of nothing but NAND gates, and led a summer research team on ultracold atomic physics. That means when I write software, I understand what's happening all the way down.
Right now I work AI-native. My day-to-day build environment runs on Claude Code with MCP connectors, version-controlled to GitHub, with human approval gates before anything touches a live system. I use that setup to design, generate, and review code end to end — while owning the requirements, integration, and verification myself.
[Add a sentence or two in your own voice here — what you care about, what you're aiming for, or what you're learning next.]
A mix of software and hardware. Each one I took from idea to working build.
An end-to-end AI build environment in VS Code using Claude Code as the primary engine, version-controlled to GitHub. Configured portable Claude Skills (reusable instruction modules) and integrated MCP connectors so agents act directly on external tools and APIs — with human approval gates before any external action.
A motorised car built from scratch on breadboard. Engineered the full drive chain — MOSFET motor drivers, potentiometer speed/steering control, flyback diodes for back-EMF protection — and conditioned the noisy signal with a Schmitt trigger. Programmed the Arduino to poll an ultrasonic sensor in real time and brake autonomously before a collision.
A fully working vending machine built using only NAND gates, exploiting functional completeness. Coin counting, item selection, and dispensing logic implemented from first principles — both combinational and sequential state-holding circuits designed by hand.
Led a student team through a summer research programme on BEC and ultracold atomic physics. Scoped the work, coordinated workloads across members, and presented the team's findings — early proof of technical leadership and clear communication.
More on GitHub — github.com/edison9733
Available for freelance and project work. If a task can be automated or handed to an AI agent, I can build it.
Built to actually work in production, not just demo. Prompt engineering, tool use, and approval gates before anything goes live.
Connect your tools and kill repetitive work. MCP connectors wiring AI into real APIs and services.
Front to back — Python, JavaScript, React, Next.js. Clean, responsive, and maintainable.
Solana, Rust, and dApp frontends. The same AI and automation edge, applied to crypto projects.
[$15]/hr · starting rate
[Adjust or remove — your call.]
I communicate clearly, move fast, and own the work from start to finish. Whether you're hiring, collaborating, or have a project in mind — get in touch.