Learn how Ray tackles the challenges of generative AI infrastructure in this 32-minute tech talk presented by Zhe Zhang, Head of Ray OSS at Anyscale. Explore how leading AI organizations like OpenAI, Cohere, and EleutherAI leverage Ray's flexibility and architectural advantages for training large language models at scale, including ChatGPT and GPT-J. Discover the platform's capabilities for both large-scale training and orchestrating open-source generative AI workloads without starting from scratch. Gain insights into Ray's multi-node training and serving capabilities through Alpa, and preview upcoming improvements designed to enhance the platform's usability. Recorded at the ML Meetup in San Francisco on April 4, 2023.