The Data Exchange Podcast: Zhe Zhang describes how companies are using Ray in large-scale production applications.
Subscribe: Apple • Android • Spotify • Stitcher • Google • RSS.
In this episode of the Data Exchange, I speak Zhe Zhang, Engineering Manager at Anyscale where he leads the team that works on the Ray and its ecosystem of libraries and partners. Ray is an open source, general purpose framework for building distributed applications (more details in this post and video).
We began our conversation by discussing how three large companies – Ant Group, Uber, and Amazon – have been using Ray:
- ❛ With regards to Ant Group, Uber, and Amazon, the thing that I really like about these three very heavy production use cases is that they’re stretching the limits of Ray in different dimensions. Uber is really pushing the limits of Ray in terms of offline machine learning, in training both classic ML and deep learning and hyperparameter tuning. Amazon is really pushing the limits of Ray for really heavy load data processing, for example, they’re accessing the object store very heavily. Ant Group is pushing the limits of Ray in terms of sheer scale and the flexibility. Ant Group is using Ray in many kinds of compute workloads that combine different environments.
We also discussed the state of the Ray ecosystem and the project’s near-term roadmap, and we previewed the upcoming Ray Summit. Ant Group, Uber, and Amazon are among the many companies who will be speaking at the conference.
Related content and resources:
- A video version of this conversation is available on our YouTube channel.
- Max Pumperla: “Connecting Reinforcement Learning to Simulation Software”
- Piero Molino: “Making deep learning accessible”
- Dean Wampler: “Scalable Machine Learning, Scalable Python, For Everyone”
- Michael Mahoney: “Tools for building robust, state-of-the-art machine learning models”
Subscribe to our Newsletter:
We also publish a popular newsletter where we share highlights from recent episodes, trends in AI / machine learning / data, and a collection of recommendations.
[Image by Dmitriy Gutarev from Pixabay.]