01 AI · LLM & Cloud
Serving open LLMs at scale — vLLM on Google Cloud Run, inference optimization, agentic pipelines.
Daejeon, Korea
At the intersection of AI, blockchain, and the physical world.
Coding + Business. I ship LLM systems on the cloud, build blockchain infrastructure, and teach machines about the physical world — then write about what I learn.
NEW Relaunching this site as a working journalA quieter, editorial home for notes on AI, blockchain, and building — with a homepage t...
Meta💬 0
Serving open LLMs on Cloud Run with vLLMNotes on getting throughput up and cost down when you serve your own models instead of ...
AI💬 0
What an L2 actually has to get rightRollups are easy to describe and hard to operate. A short list of the parts that decide...
Blockchain💬 0
Giving robots a usable model of the worldPerception tells a robot what it sees. Knowledge tells it what that means. The gap betw...
Physical AI💬 0
Multilingual batch document translation into 50+ languages.
Designed and built end-to-end — Excel, Word, HWPX, PowerPoint, PDF, and Google Workspace files translated in one pass. Preserves tables and layout, caching cuts cost up to 80%, with glossary control and TTS in 40+ languages.
Serving open LLMs at scale — vLLM on Google Cloud Run, inference optimization, agentic pipelines.
Ethereum & OP-stack L2 infrastructure with the Tokamak Network ecosystem.
Ontology-based reasoning for embodied agents and robotics (KnowRob).
Product engineering across React · TypeScript front-ends and Spring · Java services.