“A Shot in the Dark Email & the Rise of Anyscale”

By 2016 & 2017, I was heavily invested in learning about technologies in the Apache OpenSource ecosystem, primarily because it had cristened and spawned so many incredible tech companies up to that point. The fascination for open-source software led me from Cambridge, Ma to Berkeley, California where I found a pattern emerge. Many of the companies I was reading about were all founded in the confines of Soda Hall at UC Berkeley & many of which were studying under Professor Ion Stoica. The group within Soda Hall was originally called the AMPLab and included Matei Zaharia who made Spark which would later become Databricks. As I learned about Spark and other related Apache projects like Kafka, Cassandra and Mesos, I made it a point to keep tabs on new projects that were coming out of the halls of that unique building. Professor Stoica was, and still is, inspiring students in droves.

In late 2017, I saw a paper that surely caught my attention from students inside Soda Hall called Ray, a framework for distributed ML across machines that incorporated other early source software frameworks like Uber’s Michaelangelo. The lab was now called RISELab and the authors of the paper were Philipp Moritz and Robert Nishihara. I was so intrigued by the work that I sent an email to Philipp and Robert asking if people were building with Ray as of yet. Shortly after, I received an email from Philipp and then Robert who sent me the early version of the Ray paper that was under submission and pointed me to early code repos for the Open-Source project. It was thrilling to be an early reader of the paper. I had a feeling Ray would become something if they tried, if only I had millions to invest at the time.

A few years went by and I heard nothing of Ray in the community and figured it may have been put aside. But in 2019, I witnessed the launch of Anyscale by Nishihara and Moritz using open-source Ray as it’s foundational framework to support AI & ML workloads of all kinds.

By 2022, the company has reached Series C investment with it’s latest round of $99 million led by A16z and Intel Capital. Anyscale makes it effortless for companies of all kinds to run scaled out distributed ML & AI processing, tuning and inference applications and now sees companies like Linkedin, Lyft, and others using the platform. Yet another company from Professor Stoica that has hit a pain point, leveraged the new opportunities in accelerated computing, and created a business that is beginning to pick up steam. Check them out at Anyscale.io and watch out for Ray Summit 2024 later this year.

And as we venture further into a growing AI arena, everybody is asking how do we evaluate these new powerful AI Models? Professor Stoica may be proving to have the answer as his “Chatbot Arena” on Hugging Face is where people are heading to find the evaluation answers. He can’t stop, he won’t stop.