Academia / Higher Education
The University of Florida's extraordinary AI infrastructure—the supercomputer known as HiPerGator—is used to build a host of AI solutions to serve students, faculty, staff, and researchers. The results have been far-reaching across the university, especially with the family of NaviGator chatbots that bring the capabilities of large language models (LLMs) to every facet of university life. To power this vision, the university uses NVIDIA NIM™ microservices and Llama NIM on its HiPerGator supercomputer to efficiently serve and scale AI models like NVIDIA Nemotron™ within NaviGator, optimizing GPU utilization, performance, and on-premises data control.
University of FLorida
Generative AI / LLMs
University campuses will be transformed by AI. Imagine an environment where every student has a personalized assistant that follows them through college to career, providing tutoring and resources to connect to on campus. Every department has access to foundation models and AI infrastructure to perform groundbreaking research. Every professor and faculty member has AI tools to develop curriculum and plan courses, enriching the lives of their students and building the next-generation workforce.
This is the vision being implemented at the University of Florida.
“When I talk to presidents and provosts around the country, I give them a very clear message: If you don’t do AI on your campus, the students are going to vote with their feet and go somewhere where they can learn this stuff, because they know it’s important to their futures,” said Joseph Glover, provost and executive vice president of academic affairs at the University of Florida.
As one crucial step in achieving this, the University of Florida built NaviGator, one of the first self-service, multi-model AI platforms among U.S. higher education institutions. NaviGator is a suite of tools that gives students, faculty, and staff access to a wide variety of proprietary and open source models and services to accomplish a wide range of tasks, from helping people use AI with their own datasets, to running autonomous experiments, or providing students with a personalized tutor.
Florida Institute for National Security, UF
The key motivations for building NaviGator as a private, in-house platform were cost predictability and data privacy. Essentially, the university is future-proofing its infrastructure and solutions, allowing it to adapt and grow with the rapid evolution of AI.
Both NaviGator and HiPerGator are initiatives fully controlled by the university and maintained centrally through its IT department, allowing it great freedom to innovate and build new capabilities. Central administration helps control costs and which models to make available, avoiding lock-in into any AI solution, and mapping each task to the best model. For certain tasks, the university can keep data on premises with HiPerGator or use open source models only to ensure privacy, security, and compliance related to sensitive research data or student personally identifiable information (PII).
“The practical reason is that if you do it centrally, then you can control the costs,” said Glover. “We have leveraged everything that central IT does, and that’s allowed us to produce tremendous cost savings, as well as share the benefits of it with everyone in a very seamless fashion.”
NaviGator is a true hybrid service that incorporates the on-premises HiPerGator infrastructure with access to AWS, Google Cloud, and Azure. HiPerGator is one of the most powerful supercomputers and university AI factories in the United States.
Chris Malachowsky, cofounder of NVIDIA, speaks with students about their engineering projects during the unveiling ceremony of the fourth generation of HiPerGator.
Within the next several years, the hope is that the NaviGator platform will evolve from assistants into a system of AI agents that execute certain workflows autonomously for students, faculty, and staff.
“We have a philosophy of positioning ourselves for the future,” said Glover. “NaviGator Tutor is an example of an assistant that we think will likely migrate into agents.”
Eventually, the university would like to develop a personalized “AI buddy” for every student to serve as a personalized companion that stays with them throughout their entire life on campus. As each student enters the university system as a freshman, they’d receive an AI assistant to guide them through their experience at the University of Florida, learning and growing with them, connecting them to people and resources in subjects they are interested in.
“It may be a piece of software, or it may be a robot by the time we’re done,” said Glover. “Students will show up on campus, and the buddy says to them, what do you like to do? And the student may say ‘I like to play chess’ or ‘I like to dance.’ And then the buddy begins to connect them with resources, other students with similar interests, or helps them to schedule their classes. This is something that we’re quite excited about, and we have a group of people right now who are in the planning stages.”
Join innovators like the University of Florida in shaping the future of AI with NVIDIA NIM.