01In progress
Autonomous agent memory systems
What does it take for an agent to remember meaningfully across sessions? Working on persistent memory architectures that compound rather than truncate.
HakyunRyu.
Singapore·1°N
BuilderCollectorSystems Thinker
Scroll↓
Watches◦AI Builder◦Data Analytics◦Second Brain◦Capital Markets◦Horology◦Autonomous Agents◦Philosophy◦Watches◦AI Builder◦Data Analytics◦Second Brain◦Capital Markets◦Horology◦Autonomous Agents◦Philosophy◦
Note 01
Philosophy
Same instinct that got me into watches got me into AI.
I've always been drawn to systems. Financial markets, independent watchmaking, automation, knowledge networks, the way humans think and operate. The same curiosity that pulled me into investing and independent horology eventually led me deep into AI agents and autonomous workflows. Find what matters before it's mainstream, figure it out myself, then bring others along.
My work sits at the intersection of technology, finance, and design. Less hype, more about systems that are useful day to day. I'm currently exploring how autonomous AI systems, personal knowledge graphs, and intelligent workflows can reshape the way individuals and small teams operate, especially through live data pipelines and AI tooling for non-technical SME owners.
Beyond the technical side, I'm deeply interested in independent watchmaking, thoughtful product design, and how philosophy shapes the way we interact with the world. I'm especially drawn to tools that feel innately human. Quiet infrastructure that amplifies thinking rather than replacing it.
I care about building things that compound over time. Knowledge, systems, relationships, ideas.
- 01Find the thing before it's mainstream. Figure it out. Bring others with you.
- 02Build systems that are useful day to day, not impressive in a pitch deck.
- 03The interesting questions live one floor below the dashboard.
- 04Aesthetics isn't decoration. It's a form of clear thinking.
Selected Work
(06)
Work treated as considered
objects, not deliverables.
Currently Exploring
Threads of thinking
Questions I’m turning over,
not yet answers.
02In progress
Personal knowledge graphs
Beyond chat. Building structured second brains where ideas link to each other automatically, and the graph itself becomes the thinking surface.
03In progress
Human-first AI workflows
Quiet infrastructure that amplifies thinking rather than replacing it. The opposite of the chat-everywhere paradigm.
04In progress
Independent watchmaking
Studying the makers working at the edges of mechanical horology. What survives when finishing matters more than scale.
05In progress
AI-native SME tooling
Live data pipelines and AI tooling built specifically for non-technical small business owners. The underserved middle of the AI market.
06In progress
Second-brain systems
Karpathy's LLM wiki pattern, applied. An evolving knowledge architecture that gets more useful the longer it runs.
Atmosphere
(06)
The objects, tools, and rooms
the work happens inside.
Writing
Long-form notes
Thinking, in long form.
Contact
Begin a conversation.
If you're working on ambitious systems, unconventional ideas, or thoughtful products — let's connect.