notes

Short notes on research ideas, project updates, and things I am learning while building LLM and urban AI systems.

hidden-state monitoring for safer dialogue

May 20, 2026

A short note on why implicit harmfulness is hard to catch from text alone.

llm-powered digital twins for urban mobility

Apr 12, 2026

Notes on connecting simulation, traffic data, and AI agents.

what makes meme understanding difficult

Mar 8, 2026

A brief note on captions, background knowledge, and surface messages.

building practical ai systems

Feb 14, 2026

A compact note on pipelines, interfaces, and evaluation.