Database Reference
In-Depth Information
language of that world. It goes both ways. It is not about making everybody
well rounded, it is more of agreeing and learning a vocabulary to communicate
with. Once you do that, it is a huge step forward.
Gutierrez:
When you are looking to hire someone, how do you think about
the process and how does the communication aspect play into it?
Perlich:
Internally, we have been having an interesting conversation around
the idea that being a data scientist is one of the hardest jobs ever, and so how
do we properly hire one. This discussion has been ongoing as we recently
hired two people. The interview happens in two main stages. First, we give
them half an hour to make a presentation on their previous work. This pre-
sentation can be one or two things that they have done before. Then we talk
a bit about some of the problems we are working on and ask them for their
thoughts. Throughout the interview we look for four main things: smell test,
critical thinking, communication, and how they drive the process to under-
standing the data.
I look at the smell test first. I need you to have the sense of when something
in the data feels wrong. That is the thing I cannot teach you if you do not at
least have some sense. I can teach you language. What I do not need at all is
domain knowledge. When we first talked to our HR people about needing
data scientists, they asked if we needed them to know about digital advertising.
We told them, “No, absolutely not. We could not care less.” On the contrary,
I prefer somebody who has done ten different things in ten different domains
because they will have hopefully learned something new about data from each
of different places and domains. I would rather have the breadth than the
depth. So forget domain knowledge. I assume they are smart enough to learn
they need to know about the domain in two weeks to three months.
After the smell test, we look for the rest of the criteria. I do not think commu-
nication is explicitly stated in any requirement set, but it is something that we
pick up as we talk with the person. In terms of critical thinking, we may show
them plots of data and say, “Is there something that strikes you odd here?” We
want to see how they process that information and come up with questions
that they feel are important to really get to the bottom of what is going on.
Then we move on to saying, “Okay, here's a problem.” For us, even more
important than talking about the problem, is that they are able and comfort-
able to start asking the right questions. When I present a scenario, I want
them to ask, “What do you mean by that?” They need to know when they
really understand something and when they do not, so being open-minded and
being able to drive that process is very important. After all, that is what they
will have to do as data scientists here.
Gutierrez:
What problem did you explore with one of the recent hires?
