Hyperquery Community Spotlight #1 — Robert Yi, Ph.D.

Robert Yi, Ph.D.

Where are you based?

San Mateo, CA

Tell us a fun fact about yourself.

I really like mechanical keyboards.

How did you get into your field? Tell us your story.

I became a data scientist after my Ph.D. program at MIT. I ended up getting into data science when a coworker prodded me to switch from MATLAB to python, and it all went downhill from there.

How do you use data at work?

In the past, I’d say I’m most proud of the work I’ve done on uplift modeling (I open-sourced the library pylift at Wayfair) and on CUPED (I came up with some cool, novel variations to drastically reduce experiment runtimes at Airbnb). But that was a small part of my jobs in both cases — I ran product analytics, made Airflow jobs, managed experimentation, and generally spanned the gamut of data work. These days, my hands are in product and design at Hyperquery, so I spend a lot less time hands-on with data, but I do manage our data stack, from data ingestion through Stitch and Segment into our warehouse, Bigquery, as well as our dbt modeling layer and dashboarding tools on top. Much more infra-level work, but it’s been a fantastic change of pace.

What do you love most about your job?

I love finding a problem no one has solved before, and trying to solve it. Sometimes I obviously fail, but when I succeed it’s got to be the best feeling in the world. I’ve found that this sort of innovation can span across industries, so while it used to be in finding new algorithms, these days it’s more in finding novel product patterns, writing complex SQL transformations, and/or building our open source library, Whale.

What do you dislike the most about your job?

That it sucks up my entire life! The stuff I get to work on is so interesting I have a hard time ripping myself away from it. :)

What important trend in data are you most excited about?

dbt and the CLI movement. With the rise of dbt there has been a sort of resurgence in the popularity of CLI tools for data work. I’d love to see some data workflows become more and more technical

PR time: what do you want our community members to know?

I really, really like mechanical keyboards. Also, we’re hiring, I guess. ;)