Database Reference
In-Depth Information
have those systems in parallel. Maybe eventually we'll get completely away
from the data center. But it's a matter of efficiency and investment in what
works best for our needs.
Gutierrez:
How do you think about the technology selection for general use
within your team?
Smallwood:
It's a lighter thing when thinking about technology selection for
the team rather than for the data infrastructure. We want to do analysis and
then have some visualization on top of that analysis. For this work, we experi-
ment with all different types of tools, and the technology choice essentially
comes down to whatever the passion is of the person who's working on it.
We use all sorts of tools at that layer of the stack. For analytics, we heavily
use R. Any open source software for the most part is preferable to licensed
software, so we are heavily open source-oriented. We use a ton of R, Python,
and things that are easy for people to pick up, learn, and then do all sorts of
visualization things with as well.
Gutierrez:
Whose work is currently inspiring you?
Smallwood:
My colleagues are very inspiring to me, both on my own team
but also other colleagues across Netflix. We have an amazing guy leading the
algorithms from the product innovation side—Carlos Gomez-Uribe. He's an
outstanding, brilliant guy, and his whole team is super strong and inspiring. I
feel silly saying this over and over, but it really is true: We have amazing people
at Netflix!
Outside of Netflix, on the experimentation side, I always enjoy Ronny Kohavi,
who is a great speaker and excellent at experimentation. He really gets the
power of it and is great at conveying that in his talks. In terms of algorithms
and predictive models, so many people doing great stuff that it's less about
following particular people than it is about noticing particular papers or appli-
cations coming out the companies that are doing interesting things in experi-
mentation or in the predictive algorithms and models space.
I've always thought that Amazon has done and continues to do super-amazing
things with data. LinkedIn also continuously does interesting things with data.
Some of it is that they have interesting data, but they also have like a long his-
tory in data science with DJ Patel, and progressing from there. I've always been
a fan of LinkedIn and their output.
Gutierrez:
What inspired you to get involved as an advisor to HiQ Labs, a
company that does predictive modeling on employee churn?
Smallwood:
I was drawn to it really because of the application. I'm always
looking for interesting things you can do with data. To me, what they are doing
