Talking Shop: An Interview with Collin Pitmon, Director of Data Science
Talking Shop is an interview series with members of 7Park Data’s product, engineering and data science teams.
What is your role at 7Park Data?
I am a Director in our Data Science organization, leading the Data Products and Platform team.
How does your work help 7Park’s customers?
To create 7Park Data’s market leading data products, there is a ton of work behind the scenes in taking messy data from its source all the way to something that’s actionable and decision-ready for our clients. At the end of the day, the data and analytics we produce need to carry a signal that’s relevant for what our clients care about, that’s accurate and available fast enough for them to make a decision. To do this, we have to use a diverse set of statistical and machine learning techniques and apply them at a massive scale. One of the biggest parts of our job, and value-adds for our clients, is the way we enrich and link data from different sources in creative ways. The resulting sum is worth more than the parts.
What’s the most challenging project you’ve worked on at 7Park?
The most challenging project I’ve worked on in my time at 7Park has to be our reported metric estimate products derived from our Insurance POS dataset. What made it so difficult was the complexity introduced by the behavioral economics of car purchasing and the car insurance market. It was quite a challenge developing our statistical experiment design to isolate the signals we were after. There are so many intricacies that govern how and when people buy cars, and how they behave when they buy insurance for them in the separate car insurance market. It was an opportunity for me to dust off some of my old econometrics textbooks from undergrad which was, for me, a tremendous amount of fun. I am obviously a riot at parties.
A recent, fun project was working with a few colleagues on a project we called “Float Like a Butterfly, String Like A/B” for the Vista Global Hackathon. We added semantic search on top of language transcription to create a really useful tool, especially now in the work from home era. In addition to my skills as a data scientist, I am also a skilled voice over artist.
What do you believe is the biggest misconception about Data, AI or ML?
The biggest misconception about artificial intelligence is the perception versus the reality of what it can and cannot do, and the change it will bring about in the future. On one hand, you read headlines about algorithms performing better than humans on certain tasks or even games, and my favorite sensational hot take: AI is going to take all our jobs. I do believe that artificial intelligence will change our industry and have a big impact on the future, but it will be slow and steady, and in the end everyone will benefit from it.
On the other hand, which is reality, the (massive) set of technological capabilities that we refer to as “AI” are very powerful in tasks that are so narrowly defined, it’s hard to compare to human intelligence. A majority of the artificial intelligence we interact in our day-to-day lives isn’t the result of some sentient all-knowing machine; it’s a person, probably slouched over a MacBook, sacrificing blood, sweat, and more often than not, tears.
What are some of the trends or industry changes that you’re paying close attention to?
Over the past five or ten years, there has been a huge shift in the way companies think about and use data. No matter what type of goods or services companies sell or what industry they’re in, they’re also data companies due to how they use data to make better decisions, make better products and in general run a better business.
This shift is not just coming from the top in these businesses either. In today’s world, with a little bit of commitment to learn and the ability to copy and paste code from the internet, just about anybody can create analytics or predictions that are useful. The democratization of data and analytics has been exciting to watch and fun to be a part of over the past decade. One area I really enjoy watching is seeing data science and machine learning support social good, by empowering data-driven decision making in medicine, philanthropy and public policy.
What hobby have you picked up during quarantine?
Cooking has always been a hobby of mine, but quarantine really offered me the opportunity to thoroughly practice and study the fundamentals. Cooking is great because it’s a wonderful combination of both art and science. Cooking is especially enjoyable for programmers. You spend the effort preparing the ingredients, like chopping onions and you put them into the pan to sauté. Your cast iron doesn’t suddenly throw a critical error notifying you the knife you used to chop the onions is four versions out-of-date, and you need to install a package dependency that’s no longer compatible with your operating system.