Teaching AI: A Student-Built Expert System with LLMs, Graphs, and Memory

Session Track: AI Engineering

Session Time: 12:30 am - 01:00 am November 7

Session description

This talk showcases an expert system developed for use in an undergraduate business analytics course. The project integrates LLMs, Neo4j graph databases, and vector-based memory to create a dynamic, context-aware reasoning framework. By employing a multiagent architecture and retrieval-augmented generation (RAG), students designed a system capable of maintaining long-term conversational memory, generating transparent responses, and drawing insights from structured and unstructured data. The presentation highlights the pedagogical design of the project, the learning outcomes achieved, and the system's broader potential as a teaching and prototyping model for graph-powered AI solutions in applied settings.

Speaker

Joe Wilck

Faculty, Analytics and Operations, Bucknell University

Dr. Joe Wilck has been teaching analytics, operations research, data science, and engineering since 2006. His research is in the area of applied optimization and analytics, and it has been funded by the National Science Foundation, Department of Energy, Defense Advanced Research Projects Agency, and North Carolina Department of Transportation, among others. He is a registered Professional Engineer. He is an active member of the Institute of Industrial and Systems Engineers, Institute for Operations Research and the Management Sciences, and Military Operations Research Society. He is a faculty member at Bucknell University.