CGU SAGE

From request to rollout

CGU SAGE grew out of a new AI Innovation and Support Center at Chang Gung University. As founding lead, I built the center’s technical foundations from the ground up, defined its roadmap and operating model, and recruited and led three student developers.

Rather than treat each faculty request as a one-off demo, we built a shared path from an initial teaching, research, or administrative need to a deployed agent. The platform made institutional sign-in, scoped access, credential handling, usage and cost controls, and data boundaries part of that path.

My role

I led faculty projects from problem definition and implementation through testing, classroom rollout, and iteration. The work connected product design, platform architecture, and direct delivery with instructors.

I also built a coding-agent development harness around the codebase. Structured repository context, explicit capability boundaries, hooks, and CI/CD feedback helped guide and validate agent-driven changes. To maintain continuity beyond my term, I recruited and onboarded a part-time successor.

Public reference implementation

For teams that need to self-host an agent, I designed and built the CGU SAGE Agent Starter, with reviewed Node.js and FastAPI reference implementations. The public repository demonstrates the integration pattern without exposing CGU’s private platform code or operational data.

The repository is publicly viewable source governed by the CGU Internal Use License. It is not an open-source personal project.

Ownership and disclosure

CGU SAGE is an institutional project of Chang Gung University. This case study describes my role and public demonstrations. CGU’s source, product assets, and related materials remain Chang Gung University’s property.

This portfolio does not republish private source, internal data, credentials, configuration, learner information, or restricted screenshots.

View the CGU-owned Agent Starter ↗