AUSME Faculty Expertise RAG System
Retrieval-augmented search and summarisation over roughly 25,000 research papers, built to surface faculty expertise and collaboration opportunities.
Problem
Institutional research expertise is real but effectively unsearchable — distributed across thousands of PDFs with inconsistent metadata. Finding "who here works on this, and what have they shown?" is a manual task.
Architecture
A LangGraph-orchestrated ingestion pipeline: document parsing, metadata structuring and embedding generation, feeding semantic retrieval over the full corpus, with contextual summarisation grounded in retrieved documents and exposed through a conversational interface.
Dataset
Approximately 25,000 research papers, collected by scraping metadata and full texts, covering 50+ faculty members.
Results
The complete retrieval pipeline processes the entire corpus in about 30 minutes.
Future work
A proposed deployment architecture allowing faculty and research groups to query publications in natural language, improving visibility of research domains and supporting interdisciplinary team formation.