Robert Nishihara is one of the creators of the Ray, a distributed system for scaling applications to clusters. He is one of the co-founders and CEO of Anyscale, which is the company behind Ray. He did his PhD on machine learning and distributed systems in the computer science department at UC Berkeley. Before that, he majored in math at Harvard.
Robert Nishihara is the creator of the wildly popular Python project Ray for parallel ML and the founder/CEO of Anyscale.
How did you come to learn about open source? What was your introduction to open source early in your career? - 0:26
Before getting into Ray, what got you into this problem space of parallel computing distributed systems? What drew you into that world from an academic perspective out of Berkeley? - 1:22
Can you give us an overview on Ray and how it was created? - 3:01
One of the more recent projects that the RiseLab created is this open-source project called Ray. Can you give us an intro to Ray and why Ray was created? - 5:14
What are the core focus areas and scope of Ray today? (See: github.com/ray-project/ray) - 6:58
That’s a really comprehensive understanding of Ray as this sort of universal layer for machine learning in the Python ecosystem, but even more broadly. So even non-Python applications and non-Python oriented ML systems can benefit from Ray? - 10:20
What are the most exciting features you’re starting to see in Ray, and what does the short-term/medium-term roadmap look like for Ray? - 11:44
What are the interesting use cases that you see for Ray out there that are surprising or in line with what you envisioned? - 13:58
What does the ecosystem around Ray look like today? You have cloud providers, developers. How do you develop a mental model around this ecosystem? - 15:50
I’d like to get into Anyscale. You’re the co-founder of Anyscale, going from academia to creating an open-source project to being a startup founder. Why was Anyscale founded? And what was your personal journey like? - 17:57
It sounds like the company was found in the last 1-2 years, and now you’re close to 50 people. What are the most critical areas you’re hiring for as you grow the company in these next two years? - 21:00
What are the biggest challenges to scaling Ray and scaling Anyscale, today and in the future? - 23:35
Share your questions and comments below!