Scaling AI infrastructure to meet growing demand comes with major challenges around cost, control, and performance. Red Hat’s Will McGrath, James Harmison, and Philip Hayes break down three approaches for tackling these issues, harnessing the power of open source platforms and scaling for greater performance.
Learn how companies like Turkish Airlines are optimizing GPU resources across hybrid cloud environments, and how Models-as-a-Service and scalable inference tackle duplication and help achieve consistent performance for AI models in production.
00:00 Introduction
02:08 Three approaches to scale AI infrastructure
02:41 GPU-as-a-Service: Optimizing GPU resources
05:12 Demo
14:57 Models-as-a-Service
17:26 Today’s infrastructure, tomorrow’s strategy
19:30 Fast, flexible, and scalable inference
21:50 Introducing llm-d
24:20 Demo
29:33 Conclusion
🚀 Explore our AI portfolio: https://www.redhat.com/en/products/ai
#RedHat #AI #llmd
