Matheny S1, Gambarini V1
1University Of Auckland , Auckland, Aotearoa / New Zealand
Biography:
Sean is the Platforms and Services lead at the University of Auckland's Centre for eResearch. He has a has a background in radiation oncology treatment and planning digital workflows and oversees and develops eResearch workflows and platforms at the University of Auckland. He has co-developed the L.A.I.S. platform.
Victor brings his extensive experience in bioinformatics to help inform and advise researchers at University of Auckland, Centre for eResearch. In addition to co-developing the L.A.I.S. platform, he helps operate the New Zealand node of the ARDC's Nectar.
Abstract:
AI solutions for research are advancing at a breakneck pace, but most infrastructure targets massive-scale datacenters and HPC clusters. Current supply chain bottlenecks and unpredictable cloud token pricing mean we must find creative AI solutions at every tier of the research computing ecosystem.
In this spirit, we have developed a novel service leveraging NVIDIA’s compact GB10 architecture. Weighing less than a kilogram, consuming under 200W, and operating silently, these units keep data entirely on-device—a critical advantage for sensitive data. We have integrated 12 of these units into the L.A.I.S. service, allowing researchers to reserve dedicated inference units to automatically host LLMs or utilize advanced remote terminal access.
Our goal is to bridge the gap for mid-tier AI needs, ensuring that scarce, ultra-expensive enterprise clusters and HPC resources are saved for the most demanding workloads, whilst giving everyday researchers immediate, secure AI capabilities.
