CHENG-HAN LINM.S. student, CSIETaichung, TaiwanRAG · self-hosted inferenceML · backend · payments
Three shipped products, each built solo. My work is about turning large language models into something people can trust — retrieval-augmented generation so the output is traceable, self-hosted inference so sensitive data stays put, and machine learning for the forecasting an LLM is bad at, together with the backend, mobile and payment infrastructure underneath.
B.S. in Computer Science and Information Engineering, National Chin-Yi University of Technology, June 2026; on the M.S. programme in the same department from September 2026.
Capstone project
2025–2026
AI mood-journalling platform
A mental-health journalling platform built on retrieval-augmented generation, with a Random Forest emotion forecast and self-hosted open-weight model inference. Every layer of it — ML, backend, frontend, mobile, infrastructure — was built by one person.
2026 onwards
Subscription reminders that arrive before the charge — over LINE, email or desktop, not a receipt after the money has gone. It is a paid product itself, so the recurring billing and the government-mandated e-invoicing behind it are mine too (Taiwan's, though the EU, Latin America and India all run the same kind of scheme).
One inventory a household shares, with a different level of access for each member and full editing while offline. Paste a block of text from Notion, Excel or Google Sheets and it works out the items, quantities and units rather than making you retype them.
Happy to talk about research, roles, or the engineering behind any of it.
alan930604@gmail.com