Overview
Watch a comprehensive technical presentation where Dave Stiver, Group Product Manager at Google Cloud, delves into cloud storage solutions optimized for AI/ML workloads. Learn about critical components of the AI data pipeline including training, checkpoints, and inference, with a focus on developer-friendly storage interfaces. Explore key features like GCS FUSE for mounting cloud storage buckets as local file systems and Anywhere Cache for enhanced data access speeds near accelerators. Discover real-world implementation through Toyota's Woven case study, which achieved 50% cost reduction and 14% faster training times by switching to GCS FUSE. Understand the architecture and benefits of Parallel Store, a fully managed parallel file system built on DAOS technology for high-throughput AI workloads. Gain insights into managing large datasets across hybrid environments while optimizing both cost and performance for diverse organizational needs in the evolving AI landscape.
Syllabus
Google Cloud Storage for AI ML Workloads
Taught by
Tech Field Day