Database Reference
In-Depth Information
forth. I found this to be really interesting because of the way we were using
data to compliment the work being done by other parts of physics.
Sometime after the galaxy project, I went to China to look at the transaction
data of a company that's pretty much a mix of eBay and Amazon. This project
focused on how to group all of their products and categories. We worked on
doing PCA analysis to collapse the categories to make them more manageable
and to be able to get cross-classification. That was a fun project and I feel like
I didn't get enough time there.
Gutierrez:
How did you go from doing academic data science to data
science in the industry?
Smith:
Eventually, after doing these two projects and some others, the profes-
sors advised me to go out into industry to see what other people were doing
in this space. We called around and found Bitly, a link-shortening company that
provides data insights to consumers and brands. I had seen talks on machine
learning in a grad school computer science class by Hilary Mason, formerly of
Bitly and now at Accel Ventures. The professor was like, “You should look at
this. It is amazing!” So I was really excited when I started talking to Bitly.
One of the things that really spoke to me was seeing what they were doing on
their blog. The blog was really cool because a generous portion of the content
was research-oriented, as opposed to “Here's how we support the business.”
From the talk, their blog, and what they were doing, Bitly seemed like a great
it. So I joined them to do an internship and enjoyed it so much that I left my
PhD program to become an employee at Bitly.
Gutierrez:
What was the aha! moment where you decided you wanted to
pursue working with data instead of physics?
Smith:
It was definitely a slow transition, because in undergrad and grad
school I didn't want to deal with data. In undergrad I had opportunities to do
data modeling and all things that come with data modeling. Instead, I thought
that data modeling was stupid because anyone can do that and I let the oppor-
tunities pass by me.
As I got older, I realized I liked doing more computer science-related work.
I enjoy working with code, which somewhat naturally lends itself to dealing
with data. It really hit me upon getting to Bitly, where I finally thought “Oh, this
is what I need to be doing.” There was a bit of that with my other projects in
grad school, but it wasn't as clear.
In grad school I was just one person in a black box basically. I either had my
one professor I could go to talk about things or I could go to the Internet.
At Bitly I was suddenly surrounded by people who are all very excited about
data and the different things you can do with it. Now I was able to practice the
machine learning algorithms that I've learned about and see them actually do
things. It was really cool and I finally felt like I had found what I wanted to do.
