Advanced Computing Series: National Research Platform for Agentic AI: Running Inference on GPUs and GenAI Cards
Remote event
This tutorial explores how the National Research Platform (NRP) enables agentic AI development and high-performance inference using GPUs and emerging GenAI accelerator hardware. Attendees will learn how to request and utilize resources such as NVIDIA GPUs and Qualcomm Cloud AI 100 Ultra cards through an accessible, browser-based interface. Through hands-on exercises, participants will deploy open-weight models, execute inference workflows, and benchmark performance across heterogeneous hardware environments. The session highlights the integration of JupyterHub, JupyterAI, and Kubernetes to support scalable, multi-tenant AI workloads. A key focus is on agentic AI workflows, demonstrating how users can build, orchestrate, and manage intelligent agents within persistent development environments. We will also showcase practical techniques for bringing these capabilities into classrooms and research labs, even in resource-constrained settings.
Instructor
Daniel Diaz
Computational Data Scientist
Daniel Diaz is a Computational Data Scientist at the San Diego Supercomputer Center (SDSC). Daniel earned his Ph.D. in Particle Physics from Florida State University in 2020. His research background is in large-scale data analysis for particle detector experiments, where he applies machine learning and specialized computational methods to collect, process, and interpret high-volume datasets. At SDSC, he contributes to the HPC environment and primarily supports the National Research Platform (NRP), helping users run and scale workflows on NRP.