Metro Hackathon — MCP + LLM
MCP services and an LLM response layer for passenger assistance; second place at the 2025 Metro Hackathon.
MCPLLMAPI
Overview
MCP services pair with a large language model to produce passenger-assistance responses beyond a fixed FAQ.
Results
Second place at the 2025 Metro Hackathon.
Role / Contributions
MCP service architecture and implementation, plus the LLM response flow.
Personalized passenger service
Building on Taipei Metro Go, I planned and implemented MCP services connected to a large language model. Responses use passenger context and needs; route suggestions can consider time, weather, or preferences. The architecture also allows extension to fare queries, shopping-area recommendations, and staff collaboration.