Page 12 - OIT Progress Report: 2025 Edition
P. 12

Overleaf
In Fall 2024, we licensed full access to Overleaf
Professional, a powerful online LaTeX editor
designed to simplify the writing, editing, and
publication of scientific and technical documents.
Offered in collaboration with the Faculty
Technology Council and supported through the
University’s research investment, this campus-
wide license gives students, faculty, and staff
direct access to a tool long favored in STEM and
data-intensive disciplines.
Chat_RAG
OIT has successfully developed and deployed
Chat_RAG, a secure chatbot solution with
a Retrieval-Augmented Generation (RAG)
component for sensitive and private data querying.
This tool leverages open-source LLMs running
via Ollama and the LangChain framework, fully
hosted on SMU’s ManeFrame 3 (M3) through the
Open OnDemand portal. By utilizing M3’s GPUs,
Chat_RAG eliminates the need for external API
calls, ensuring that institutional data remains safe,
secure, and processed entirely on-premise. This
makes HPC access self-service and thus easier for
Faculty to do research. This innovative solution
was presented at the PEARC 2025 Conference,
highlighting SMU’s advancements in secure AI-
driven research support.
AI Transcription Application
In early 2025, OIT introduced a research-focused
Kubernetes environment designed to give campus
researchers a secure, containerized space to
experiment with deployable applications. The
first successful launch was an AI transcription
application developed by Eric Long, a President’s
Scholar and member of our Research STARs
9 OIT PROGRESS REPORT 2024
WHAT IS ?
LaTeX is a document
preparation system
widely used by scientists,
mathematicians, and
engineers for creating
high-quality typeset
documents. It excels
at producing complex
mathematical expressions,
well-structured documents,
and ensuring consistency
in formatting. Whether
you’re writing a journal
article, a textbook, or
your dissertation, LaTeX
offers the precision and
control needed to produce
professional-quality work.
program. The goal was to be able to transcribe
lecture notes quickly and efficiently without
the need of an off-campus service. While still
in testing, early users have been able to save
significant time transcribing voice notes and
captioning lecture videos.





































   10   11   12   13   14