Ray Summit is a conference held in August of 2022 for developers, researchers, and engineers to learn and train with Ray, an open-source Python framework for building distributed applications, including scalable artificial intelligence and machine learning applications. The two-day conference allows attendees to learn about best practices, real-world case studies, and the latest research in systems built on Ray. And it brings together users and developers interested in building applications for e-commerce, media, logistics, transportation, finance, and internet of things (IoT), among others.
The Ray Summit is intended to provide an opportunity for users of Ray to build their skills, build a professional network, learn what is on the Ray roadmap, and be involved with the builders and maintainers of Ray and its libraries. This is as, in 2022, Ray is considered to extend to artificial intelligence assemblage to bring scalability to a learning infrastructure. The summit partially presents this to users of Ray, while also learning how organizations use Ray's AI application and help them develop new AI initiatives through Ray.
Other features of the Ray Summit include the following:
- Presentations on the latest developments in Ray and Ray libraries
- Learning about Ray use cases from engineers and researchers from companies including Lyft, IBM, Riot Games, and Verizon
- Presentations on teams from Spotify, Meta, and Amazon, among others, and their machine learning platforms built on Ray
- Hands-on Ray training sessions
Keynote speakers at the Ray Summit event include the following:
- Smitha Shyam, director of engineering at Uber
- Soumith Chintala, engineeer at Meta AI
- Anca Dragan, assistant professor at UC Berkeley
- Dario Gil, senior vice president and director of research at IBM
- Kim Hazelwood, engineering director at Meta AI
- Greg Brockman, CTO and cofounder at OpenAI
- Ion Stoica, cofounder, executive chairman, and president at Anyscale and professor at UC Berkeley
- Robert Nishihara, cofunder and CEO at Anyscale
Below are some of the highlighted talks of the Ray Summit:
- "Scaling AI workloads with Ray," presented by Richard Liaw and Eric Liang
- "What's new in Ray Serve," presented by Edward Oakes and Simon Mo
- "A Kubernetes Ray clustering solution," presented by Ali Kanso and Jiaxin Shan
- "What's new in RLlib," presented by Sven Mika and Jun Gong
- "Large Scale Data Shuffle in Ray with Exoshuffle," presented by Stephanie Wang and Jiajun Yao
- "Deep dive into data ingest with AIR + Datasets," presented by Clark Zinzow
- "Ray Observability: Present and Future," presented by SangBin Cho and Ricky Xu
- "The Magic of Merlin: Shopify's new machine learning platform," presented by Isaac Vidas
- "Ray + Weights & Biases: Build and deploy real-world ML models," presented by Lukas Biewald
- "Large-scale distributed training with TorchX and Ray," presented by Mark Saroufim
- "Large-scale deep learning training and tuning with Ray at Uber," presented by Xu Ning, Michael Mui, and Di Yu.
The Ray Summit is presented by Anyscale, the company behind Ray, and is sponsored by Ant Group, IBM, Intel, Weights & Biases, Amazon Web Services, ByteDance, Union, and Redis. And community partners of Ray Summit include AI Austria, Big Data Institute, Chug, Gradient Flow, Blavatnik Interdisciplinary Cyber Research Center, IGTCloud, Packt, PyData, Riselab at UC Berkeley, and Yow!.
