Introduction to Ray Data - Processing Large Datasets on Heterogeneous Clusters

Scaling AI and LLM Workloads with Ray Data: A Two-Part Technical Webinar Series

Efficient large-scale data processing and inference are critical for modern AI workloads. In this microseries, we dive deep into Ray Data and Ray Data LLM, exploring how it enables distributed data processing and LLM batch inference workloads across heterogeneous clusters.

Webinar 1 (this webinar) introduces Ray Data, covering its architecture and core concepts, usage for distributed data processing, data loading, transformation, and writing at scale, key operations like shuffling, grouping, and aggregation, as well as best practices and real-world use cases in production.

In Webinar 2, we explore Ray Data LLM. New Ray capability for efficient batch inference for LLMs, integrating with inference engines like vLLM, using a Processor-based abstraction to configure model execution, parallelism, and optimization strategies for large-scale workloads.

By the end of the first webinar, you’ll understand how to leverage Ray Data for distributed data processing and inference. You’ll leave with hands-on notebooks and a deep understanding of Ray Data’s capabilities in production ML pipelines.

LinkJoin this webinar to learn more about:

• Ray AI Libraries and Ray Data for distributed data processing

• Ray Data fundamentals: loading, transforming, and writing data at scale

• Scaling batch inference, working with heterogeneous clusters

• Best practices and real-world use cases

LinkWho Should Attend?

Ray Technical Webinars are ideal for software engineers and ML practitioners who want to jump start learning about Anyscale and Ray for large scale ML workloads. It is also useful for Infrastructure Engineers who plan to support Ray use cases in their organizations.

Don't miss this opportunity to gain deep technical insights and practical knowledge from industry experts.