Database Reference
In-Depth Information
Most of us recognize and laugh at the parable of a drunk looking for his keys
under a streetlight because that's where the light is. But we do it all the time.
We, as an industry, work with the data we have on hand and optimize what
we can measure. That's not an entirely bad thing. It's much better than trying
to work without data or trying to improve things we can't measure.
Still, a little bit of humility goes a long way. If our data tells us something that
seems incredible, the correct response is skepticism. After all, incredible is
Latin for “not to be believed.”
Gutierrez:
How does technology selection factor into solving problems?
Tunkelang:
Technology is obviously important and choosing a technology
stack is one of the biggest decisions that you make as a software engineer or
data scientist. The wrong technology selection can be a major impediment, as
it often leads to kludgey workarounds.
Technology selection by itself is unlikely to solve any problems. Technology is
like exercise equipment in that buying the fanciest equipment won't get you
in shape unless you take advantage of it. So always put talent before technol-
ogy. Get the right team of scientists and engineers, and then make sure the
technology doesn't get in their way.
Gutierrez:
What do you look for when hiring people?
Tunkelang:
I look for three things in a candidate. First, they need to be smart,
creative problem solvers who not only have analytical skills but also know
how and when to apply them. Second, they have to be implementers and
show that they have both the ability and passion to build solutions using the
appropriate tools. Third, they have to have enough product sense, whether it
comes from instinct or experience, to navigate in the problem space they'll be
working in and ask the right questions.
Gutierrez:
What do you mean by “product sense,” and why is it important?
Tunkelang:
By “product sense,” I mean the ability to see real-world problems
from the perspectives of users and other stakeholders. For example, a computer
scientist might come up with a system that improves through positive and negative
feedback. But someone with product sense would think about what would moti-
vate the users to provide the system with such feedback. On the business side,
someone with product sense will use that sense to inform key business metrics—
for example, determining when a recommendation system makes suggestions so
bad that they incur a cost beyond the user simply not clicking on them.
Product sense is a critical skill for a data scientist. Without product sense, you
can be a great software engineer and a great statistician, but it's unlikely you've
identified the right problems to solve or picked the right metrics for evaluating
your solutions. Finally, product sense can help you find shortcuts, such as getting
users to help you solve your problems.
