- In the past year, the open source project Ray, which was created by the cofounders of $95.6 million startup Anyscale to run powerful AI applications on distributed computers, has grown substantially in contributions, its community, and popularity.
- Intel now uses Ray for its own internal AI and big data applications to help customers in turn.
- Burger King uses Ray and Intel’s open source project Analytics Zoo to power recommendations in its app.
- Tencent Cloud, the Chinese cloud giant that also builds WeChat, uses Ray and Analytics Zoo for forecasting, modeling, and predictions based on its data.
- Sign up here to receive updates on all things Innovation Inc.
Anyscale, founded last year by a team of researchers at UC Berkeley, has seen rapid growth in the past year. That growth has led it to a $95.6 million valuation and a partnership with Intel, with the two teaming up to serve major customers like Burger King and Tencent Cloud.
Anyscale’s cofounders are known as the founders of Ray, an open source software project that makes it easier for developers to build applications that are distributed across multiple computers — useful for machine learning tasks that require more memory and overall processing power than any single processor can handle.
As for Intel, it had been developing an open source big data and AI project called Analytics Zoo. But its engineers saw the potential of Ray, and started using it to build advanced AI-powered applications.
It was a natural fit, in some ways: Intel already has an active collaboration with UC Berkeley’s RISELab, home to the Ray development team. In fact, Intel used Apache Spark, an open source data analytics tool borne of UC Berkeley’s AMPLab, to build much of Analytics Zoo. And so, Intel approached both labs to figure out how to bring Analytics Zoo and Ray together.
“We immediately jumped to this chance given our collaboration history and the popularity of Analytics Zoo,” Ion Stoica, Anyscale cofounder and executive chairman and a computer science professor at UC Berkeley, told Business Insider.
The Intel/Anyscale partnership comes as Ray itself sees a boom in popularity. Ray has doubled in users over the last year, Stoica says, and it now has 13,000 stars, or likes, on code repository GitHub, establishing itself as both popular and well-loved in the open source community.
Anyscale’s star has risen, too: Late last year, Anyscale announced that it had raised $20.6 million in a funding round led by Andreessen Horowitz. In September, Anyscale released a major new version of Ray, stoking the enthusiasm around the software.
“If anything, the need for distributed languages and these workloads got much much stronger now,” Stoica said. “What happened with Ray last year, the popularity of Ray increased a lot.”
All this helped set the stage for Intel to work with Ray to help its customers.
Working with Burger King and Tencent Cloud
For Intel, the appeal of Ray is that its distributed approach means that it can run more powerful AI applications, including personalized user recommendations. Ray also has a certain appeal to developers, thanks to its support for popular AI software frameworks like Google’s TensorFlow and Facebook’s PyTorch, lending flexibility in the engineering process.
All of that computing power is being put to work in Burger King’s mobile app, which uses Intel and Anyscale technology to recommend meals to customers based on their history. To make it all work, the app needs to process historic transaction data from the user to figure out what the customer might be in the mood for, and when.
This can pose some challenges. For example, customers’ preferences can change depending on where they are, what time it is, and the current weather conditions. Customers usually don’t buy frozen drinks on rainy days or kid’s meals at midnight, Stoica said.
With Ray, Spark, and Analytics Zoo, this app can take into all these factors to provide “context-aware” recommendations to customers.
“That’s why we are excited about our mission and very excited to collaborate with other people,” Stoica said. “We’re supporting more and more of these sophisticated AI applications.”
Intel is also working with Tencent Cloud, the cloud computing arm of Chinese tech titan Tencent, known as the creator of the the popular chat and commerce app WeChat. Tencent Cloud uses AutoML, a prediction toolkit for developers that’s built on top of Ray, and that’s also part of Analytics Zoo.
With AutoML, Tencent can make forecasts and build models based on data observed sequentially in time, such as analyzing network quality, gathering data on manufacturing equipment at factories, and the like.
“Building machine learning applications takes a lot of effort,” Jason Dai, CTO of big data technologies at Intel, told Business Insider. “We try to build a more general tool that can help you automate the process.”
Got a tip? Contact this reporter via email at email@example.com, Signal at 646.376.6106, Telegram at @rosaliechan, or Twitter DM at @rosaliechan17. (PR pitches by email only, please.) Other types of secure messaging available upon request.
To Find More Information, Go To Saubio Digital And Look Up Any Topic