LLMs for Transportation and Mobility

language-model representations, agentic mobility generation, and LLM-empowered transportation systems.

I am interested in how large language models can become a useful modality for transportation systems rather than only a text interface.

This includes geospatial representations from LLMs, LLM-empowered demand prediction, macro-aligned micro-record synthesis, and agentic systems for interpretable mobility generation.

Representative work: LLM4TR, LLM2GeoVec, LLMSynthor, and MobEvolve.