At Interview Query, we love talking to our success stories. This week I talked to Alex who recently joined NetworkNext as a data scientist. We talked about his challenging interviewing experiences, advice to other job-seekers, and the non-traditional ways to get into data science. You can follow Alex on Linkedin.
Hello! What’s your background, and how did you get into data science?
Hi, my name is Alex. My background is a Master's in Computational Physics. When I started, data science wasn’t so big yet in 2015, but it started to pick up quickly. I tried to get my foot in data science after I graduated but found out it was really hard. I ended up getting accepted at a bootcamp that helped me understand machine learning fundamentals and algorithms.
I went a really non-traditional route and ended up doing freelance work on Upwork and found a client, Komodo Technologies, that was impressed with some of my initial work. After a few months as a contractor, I got an offer with Statusquota, the CEO's other company she helped work on.
I've heard finding contract jobs on Upwork is difficult. How was your experience?
It definitely started out as doing odd part time jobs. There would be work sporadically for different clients but one day I contacted someone that needed basic modeling work. It ended up being Komodo when they had a posting and needed a contractor, and it made a pretty good way to go into a more consistent gig. I worked as a consultant for about a year until they ran out of resources.
I think trying to gain experience on Upwork is useful since a lot of junior data scientists don’t have experience working with clients and figuring out what a business really wants at the end. When I worked a few jobs in freelance, it gave me the experience of understanding what different businesses needed.
What was your data science interview experience like?
I started getting interviews in Seattle, Florida, and some other places I was targeting to move. I had a strong background in machine learning fundamentals and coding, but had trouble dealing with all the different types of data science interviews. Generally if I didn’t do well, I knew what I messed up on.
Some interviews were somewhat similar like Geico or Statefarm because they were both in the insurance industry. Such as there were problems that were insurance industry specific around solving problems about inequality of where data existed.
Was there an interview that was super challenging?
I would have to say that the NSA position for data science was the most memorable. You get offered to take a data science exam and you end up going into a testing facility where you take an exam at a computer with super tight security, no phones allowed or anything.
Eventually after you leave, you get an email if you passed or not and then they say they'll contact you for an interview for within months. I got the email saying I passed but then I never heard back. The exam was helpful though for showing me what concepts I needed to study for and it filled in a lot of gaps.
The most challenging interview question I had was from an interview with Karat. I ended up making it to the technical interview but then they switched gears and told me that they were looking for more of a data engineer right now instead of a data scientist. This seemed like a pretty common problem, as you can't really do data science until you build out the data infrastructure. The take home assignment was super messy data. It was a lot of exploration in trying to build a good model from messy data.
Did you find any valuable resources for your interviews?
I found Interview Query Premium really useful to me in terms of showing what the questions from other data science tech companies were like. It was really easy to go down the list and ask myself each question and made sure I knew the right answer by checking the solutions as well. I found the situational business questions helpful because the bigger companies like Facebook, Amazon, Google, etc.. are always looking for a specific person with industry knowledge.
I also used a few books. One of them was a practical statistics book by O’reilly which really helped to refresh my memory. A few of the take-homes from companies I applied to and Interview Query had model building where you needed a score at the end to pass. This actually gave me an opportunity to learn more about different boosting algorithms.
What kind of advice would you give a data scientist looking for a job now?
Don’t get bummed. Each company is different and as long as you can takeaway what you learned from the interview, then the next interview will be better. I started looking for a job during covid-19 and was in the final interview process for two companies, and they both had hiring freezes.
My rental lease was ending at the end of April so I was getting nervous that I wouldn't find a place to move to. I began to start my search all over again in Mid-March and tried to change it up to the type of companies I applied to instead of general consulting companies to traditional tech companies.
I found that talking to consulting companies was really strange given the technical screens occurred with consultants instead of data scientists. This resulted in weird case studies.
I ended up finding that the last three companies I interviewed at were my best interviews. They also ended up being the least structured data scientist interviews. But when I had an hour and half conversation with a data scientist which was very casual, I really enjoyed it a lot more since we could discuss the job and responsibilities.
I ended up taking an offer at NetworkNext. The interview consisted of two thirty minute interviews. And they just knew they wanted to give me an offer since they say my enthusiasm for working in the video game industry which has always been my dream job.
What do you think the future of data science interviewing holds?
I think the future of data science interviews need to stray away from take-homes and quizzing people. I don’t think take-homes do anything because they give user practice, but you’re doing a lot of work for not a lot back. When you can just talk to someone about a project, talk about the details, a lot more will reveal themselves.