MCP Client-Server Orchestration for Polymer Property Prediction

Designed an intelligent polymer AI Agent integrating LLMs for natural language query parsing with the TransPolymer, a pretrained transformer model for SMILES sequence sanity checking, and iterative modification by leveraging FastMCP server and finetuned Molecule Chef (2019) for suggesting candidate polymers.

Enabled automated prediction of polymer properties (bandgap, conductivity, dielectric constant, etc.) directly from polymer SMILES vector representations, and further modification of the polymer structure according to the user query.