Notes on how I work
I'm Hankyeol Kyung, an AI/agent engineer building practical systems for memory, runtime, and human-AI collaboration, from product surfaces to the infrastructure underneath them.
I'm on the product team at Asleep, where I work on SleepHub, an ambient AI product for autonomous sleep coaching and tracking. My work tends to sit where applied AI meets product infrastructure, deployment reality, and the systems that make AI useful outside of staged demos.
I care about runtime correctness, context continuity, and making AI collaboration surfaces that feel real rather than demo-shaped. Most of my work sits at the intersection of applied AI, product infrastructure, and validation, the kind of work that has to ship, stay running, and survive real usage.
I prefer explicit over implicit, inspectable over opaque, and boring technology I can debug over novelty I can't. Markdown files beat opaque stores. Plain scripts beat frameworks. When something feels magical, I usually want to see the seams.
I write about agents, memory, runtime behavior, and the architecture of human-AI collaboration on Writing. I also maintain a personal Knowledge OS that this site is generated from, a markdown-first system for turning private context into reusable task context and selected public artifacts.
The easiest public surfaces are GitHub and X. If you're working on agent runtimes, memory systems, ambient AI, or anything in that neighborhood, I'd enjoy hearing from you.