Getting to Know David Dai

Our team at Petuum is made up of incredibly talented people. This series will feature the engineers, managers, and creators that keep our company moving forward and make us proud of the work we do together.

David Dai is a product and engineering leader with a strong technical expertise in AI and ML systems, currently working at Petuum, Inc. as Senior Director of Engineering. His research spans scalable machine learning algorithms and systems, deep learning for medical imaging, and acoustic modeling.

David is currently a Ph.D. candidate in Machine Learning, Computer Science at Carnegie Mellon University. He earned a B.S in computer science from Caltech and a B.A. in physics and mathematics from Wesleyan University. He has interned at Google, Facebook, LinkedIn, Bosch, and OpenX.

Catch up with David on Twitter @daiwei89.

Why did you join Petuum?

I joined CMU in 2012 and started working with Dr. Eric Xing (Petuum’s CEO) on large-scale machine learning (ML). I have broad interests and always love to work at the intersection of different fields. Scaling ML is at the intersection of ML and distributed systems and, therefore, was a perfect problem for me. We set out to build ML technology that outperforms existing AI systems and make large-scale AI accessible to users with limited computing resources.

Through our research we quickly discovered that distributed systems designed with a deeper ML understanding can often lead to 10x or more improvements on ML problems over most existing alternatives that do not have this “ML edge”. As our research gained recognition, we wanted to bring our insight and systems to the industry to change the way people do large-scale ML. Petuum provided the crucial resources to realize this vision and thus was an easy decision for me.

What do you do at Petuum?

This is a difficult question! Because our company is growing so fast, my role actually changes over time. At first I was the release manager, making sure all parts of our product were delivered on time and functioning well. Then, I found myself making manpower plans across our product teams and building processes for inter-team communication. Recruiting is also always a big focus as we are quickly expanding our products and teams. The opportunity to talk to people from diverse backgrounds and ask them all sorts of questions is one of my favorite parts of the job!

David in Burano, Italy in 2017.

What’s the most exciting project you’ve worked on at Petuum?

The most exciting thing has been building Petuum Symphony from the ground up. We are bringing to life something that has never existed before by combining new AI technologies and UX designs. It is also the first project that we’re tackling as a team, so it definitely feels a bit like the startup experience of “jumping off the cliff and assembling an airplane on the way down.

We don’t have everything together yet, but we are figuring it out fast, and that’s part of the fun!

What’s been most challenging in your work so far?

One of the most challenging things I’ve worked on is handling our company’s growth. Every time we think we’ve figured out some organization and processes that work, we find ourselves in need of further adjustments. We are kind of in a “permanent beta” mode, and I hope that it always stays that way even as we grow to become a bigger company.

Do you have any advice for people interested in pursuing similar work?

There’s a lot of hype around AI these days and it’s important to not get carried away by it. To get involved in this work, you need to get past the smoke and actually understand AI’s capabilities and limitations. Instead of just reading the news, it’s much more productive to develop a deeper understanding of an AI application (such as how YouTube makes recommendations) or an AI technology (such as deep learning for computer vision). Since AI is a fast moving field, there is no up-to-date textbook to cover everything, so reading research papers is a must. The good thing is that AI is such an applicable tool that you can often learn it by doing it. For example, you can build an AI for your favorite board games, or even build a crypto-currency trading bot (but maybe don’t trade with real money).

David jamming with friends at his place in 2017.

What do you love to do outside of work?

I love to play the keyboard and guitar while jamming with friends or performing at my church.

I also like to swim. I used to run and play basketball, squash, etc., but swimming is more kind to my body and is weatherproof, so I’ve stuck with it. I also like reading non-fiction. Some of my favorite reads from 2017 are “Better” by Atul Gawande, “High Output Management” by Andy Grove, and “How does Sanctification Work” by David Powlison.

We’re hiring! We’re growing quickly and looking for highly trained and talented technical staff, including engineers, architects, and more. Contact us at