The Yhat Blog


machine learning, data science, engineering


"How I Interview Data Scientists" with Matt Fornito

by Matt Fornito |


This interview is featured in Springboard's guide to data science interviews. Matt is also a mentor for Springboard's Data Science Career Track, the first data science bootcamp to guarantee a job after graduation.

About Matt: Matt is the President & Founder of Summit Analytics. He has over ten years of experience as a data scientist, mentor and interviewer, with experience at Target, SportsAuthority, and Dish Network.

What do you look for when you’re hiring candidates?

I feel most comfortable hiring people with a strong quantitative background who can learn programming rather than the other way around. A Masters or a PH.D is very important to me, as I feel that undergrad is not a strong signal of success; it’s a relative breeze for most people. I prefer hiring people able to pick up programming and effective communication-- knowing and understanding what the technical problems are to implementing a solution and being able to communicate those concepts is key. What differentiates data scientists and data analysts is the ability of data scientists to deeply understand data problems and how to solve for them.

"I tend to try to hire people with STEM backgrounds, [though] it isn't a strict limitation.

I like recruiting Masters and PhDs from math and statistics, chemistry, physics, and bioinformatics and engineering. There are a small handful of people in MBAs that have worked out great for me. I am actually a PhD in organizational psychology, so though I tend to try to hire people with STEM backgrounds, it isn’t a strict limitation.

What’s the best piece of advice you can give to people going through the data science interview process?

Recruiters look at education level and the last two jobs on the CV and their pedigree. HRs only take a very quick glance at CVs, so you have to stand out in a matter of seconds. One piece of advice: get yourself into a big company that has a pedigree like Facebook, or go into a startup and take a high position so that you can stand out easily for advanced data science roles.

"Successful interviewees have a comfortable grasp on what they've worked on and are ready to tell a story..."

“Walk me through a project” questions where a hiring manager will ask exactly how you built something in the past are huge--everything from what data was used, what tools were used, what the outcomes were are important to recount clearly. Successful interviewees have a comfortable grasp on what they’ve worked on and are ready to storytell on that element and relate how their work impacted the business they were working for.

What are you testing for?

Questions I ask involve a project that tests problem solving and communication skills across the interview. I am also assessing a candidate’s passion for the company and data science. A drive for continuous learning and love of problem solving are key differentiators. Then on the technical side, I am interested in seeing candidates work on how to optimize data with Hadoop and Spark and working on the tradeoffs between different data science solutions. Do they think like a data scientist? Have they done data science work? These are important questions I am looking to uncover with my interview process.

"The key thing I am testing for is the skill of adaptability.""

I will then go into math questions such as asking how gradient descent, statistical techniques, and random forest work. A couple of situational questions where the candidate is put through a hypothetical client situation are deployed to see how the candidate would handle interfacing with clients. I have a strict requirement of ability to program in Python or R, but I am flexible with C++ and Java. I don’t believe in HackerRank like testing situations where you are expected to trace out a solution; I would rather test for adaption to new programming languages and an ability to learn skills rapidly. Anybody hired is going to have to have the latent skill of adaptability, and that is the key thing I am testing for.

Interested in a career in data science? Find out more about Springboard's Data Science Career Track bootcamp.



Our Products


Rodeo: a native Python editor built for doing data science on your desktop.

Download it now!

ScienceOps: deploy predictive models in production applications without IT.

Learn More

Yhat (pronounced Y-hat) provides data science solutions that let data scientists deploy and integrate predictive models into applications without IT or custom coding.