An industrial engineering background, now spent on two questions: how to give a language model durable context about one person, and whether the result actually holds up when a person uses it. The first is engineering, the second is research, and I would rather not do either one alone.
Currently at UXC Lab, Kyung Hee University, working on LLM-based HCI.
|
An LLM keeps the vault, you keep the thinking. A tree of linked Markdown pages that Claude Code or Codex grows, prunes, and answers from.
|
A field guide to building a model-neutral agent harness on your own machine, one that keeps improving itself instead of being rebuilt every season.
|
|
The vault as a desktop app. Tauri 2 and Rust underneath, signed builds and an auto-updater feed on the public side.
|
Quantitative HCI work on how people read, trust, and hand work to language models, plus the writing and rebuttals that come with it.
|
Languages
Product
AI & Research
