Image source

Over the years I’ve attended a lot of career fairs and company recruiting presentations, usually on a college campus. At every one of these hiring events, somebody would ask me something along the lines of “how do get a job in data science even though my major is (fill in the blank)?” It’s an excellent question and something I happen to know a little bit about since my academic background is in bioengineering.

First off, most data scientists I’ve met (until very recently) didn’t major in data science. Many of the best ones I’ve worked with majored in physics. I’ve also seen applied math, econometrics, electrical engineering, industrial engineering, and others. It’s really more about how you think and solve problems rather than how many fancy algorithms you know. For those of you in this situation trying to get your foot in the door towards a data science career, here are some some tips for every phase of the interview process.

Resume

Even in today’s LinkedIN world, the good ol’ 1-page resume still plays a huge role in the hiring process. When I review a resume, I look for the following

  • Data science classes
    • Have you taken or signed up for a data science classes or boot-camp / certificate / nano-degree?
  • Data science extra-cirriculars
    • Are you a part of the data science club or meet-up on campus?
  • Data science competitions
    • Kaggle, KDD, etc. How did you rank? Did you compete in a team or individually?
  • Excellence in your field
    • Are you all-in in your major? Maybe you have a high GPA, did undergraduate research, or published some papers in your field. Maybe you were a math champion in your home country.

If you don’t have any of these points and just verbally tell me “but I really want to get started in data science” then that doesn’t sound like you’ve taken any initiative.

Career Fair Chat / Phone Interview

My high school calculus teacher used to mock us when we couldn’t answer questions by suggesting the answer “TTMS” (teacher told me so).

Since your resume shows you have some interest in data science, I want to gauge just how interested you are. As we go through your resume and talk about the data science points, I’m looking for queues such as:

  • Are you energized by those conversations?
  • Can you explain why you’re looking for a job in data science instead of your field of study?
    • I know the real answer might be “because it makes more money.” But ask any successful person in any field and they will tell you that success requires passion. You might be able to power through 4 years of college using some discipline and Red Bull but over the span of a 30 year career, you’ll be doing yourself a huge favor by being in a job you actually enjoy.
  • Can you discuss the ins and outs of your data science resume points? Or did you just fill it with buzzwords you only have a shallow notion of?
  • Have you talked with some friends who recently graduated and are working in data science to learn what cool things they’re working on?
  • Can you go on and on about data science or do you just have 1 or 2 prepared sentences followed by silence?
  • Do your responses sound like “the right answer” from a multiple choice quiz or do you seem to be formulating original answers and can easily handle follow-up questions?
  • Can you discuss why certain steps were taken in your projects? Why a certain algorithm was chosen? My high school calculus teacher used to mock us when we couldn’t answer questions by suggesting the answer “TTMS” (teacher told me so). If your answers sound like TTMS, I’ll be thinking more about my teenage years rather than your application.
  • Can you discuss the steps taken during your data science competitions/ class projects other than the algorithm? Do you appreciate the hard work behind data understanding, data cleaning, feature engineering, and model evaluation? This is the major differentiator between class projects and real projects. In class projects, the data is often given to you neat and clean.

For the phone interview, I’ll also add some simple math problems, coding questions, and SQL queries. These are always simple to see if you meet minimum requirements and if not, how you start trying to figure things out. There’ll be plenty of time to explore the depths of your ability during the on-site.

On-site Interview

Congratulations! You’ve survived 2 screening steps (resume and phone interview) and made a good enough impression that the hiring manager has decided to allocate precious time from the entire team to interview you on site!

But first, a digression

These things take a long time not just for you but also for the interviewers. Here’s the minimum amount of time each of the interviewers need to invest:

  • Read your resume (30 min)
  • Attend your presentation (1 hour)
  • Interview you (1 hour)
  • Debrief with the team (1 hour)

That’s at least 3 hours times about 5 interviewers. Since almost everybody will be a well paid computer or data scientist, making around $60 / hour that’s: $60/hour x 3 hours x 5 people = $900 people costs. On top of that, you might be provided a meal, plane tickets, and hotel stay. That works out to about $1,000 – $2,000 to get you on site. So you can be sure we hope you’re the one we hire so we don’t have to keep repeating the process.

Face to Face Interview

You have internal standards that often exceed your manager’s expectations

So I very quickly scanned your resume right before the interview because I’m quite busy with my day job (and if it’s longer than 1 page, you can be sure I didn’t read past page 1 so I hope you didn’t put anything important in those later pages) and now we’re sitting face to face. During the on-site interview, no matter what your major, I’m giving you a similar interview and similar evaluation criteria. The reason I only scanned your resume is I’m really more interested in how you think, the depth of your knowledge, and how you communicate. I won’t ask any questions that require special knowledge. Here’s what I’m looking for:

  • I want somebody who knows basic concepts around data science
  • Why do we use data science at all? Are you just a hammer seeing everything as a nail? What is overfitting? If you have big data and major processing power, why not just use a giant look-up table? One hundred years from now, what types of things would a human still be better at than a computer?
    • This type of material is usually covered in chapter 1 of a data science book. It’s the type of material that isn’t likely to show up on an exam because it’s more conceptual and philosophical (and what kind of college technical professor would ask a question that requires an essay to answer!).
  • I want somebody who knows at least 1 algorithm well
    • I understand that if you don’t know an algorithm today, you can learn it. What makes a good data scientist is not how many algorithms you know. Most algorithm details are hidden beneath APIs anyway and that the actual implementation is just a one-line function call. But here’s what your knowledge of one algorithm tells me:
      • How well you know one algorithm says a lot about how deeply you will want study other ones.
      • Since many algorithms are similar, expertise in one can be applied to understanding another
      • Knowing important concepts in one algorithm let’s you know to find the same concepts in others.
  • I want somebody who can solve basic logic and math problems.
    • This shows your ability to think critically and organize complex issues. It will be needed for you to be able to follow along and contribute when the team is thinking through the subtleties of the modeling process.
  • I want somebody who shows passion and drive to be great
    • To do good work, it’s enough that you’ve performed the tasks somebody else told you to do. But to do GREAT work, you have to want your work to be great. You have internal standards that often exceed your manager’s expectations. You want things to be done the right way even if it means more work on your end. You comment your code and follow coding conventions because doing so makes your code not just more readable but also more professional.
  • I want somebody who I can have a technical debate with.
    • Once we’re on the same team, opinions aren’t weighted by title or years of experience; rather, the are judged on merit and team buy-in. You’ll be expected to voice your concerns and opinions. When that happens, I may not agree with you initially so we’ll have a technical debate. This happens all the time. It’s easier to describe reasons I canNOT debate with you so here they are:
      • I can’t understand you in general (no matter what the topic)
      • You can’t understand me in general
      • When I ask you a question, you answer a slightly different question
      • You go on and on about other related topics without addressing the core point of debate
      • You’re overly confident in areas where you lack successful real world experience
  • It would be nice if I also got along with you on a personal level
    • This shouldn’t be a requirement, but as human beings it sure is a bonus if we like each other. For example, if we each lunch together often, that’s more time we can spend discussing side aspects of the project or how we feel about the project in an informal setting. We can discuss things you’ve thought about it but haven’t had a chance to bring up during meetings

The Bottom Line

When all the interviewing is done, the hiring manager will go around asking all the interviewees for a simple yes/no answer. Most of the times, opinions aren’t binary so I end up giving 0-5 “stars” just like a product review. If I’m borderline I’ll just ask myself “does this candidate have the passion to be great?” The number one factor that determines success is passion. With passion, you’ll have the energy to work hard, get creative, learn from others, and produce great work. If you came from humble beginnings and worked your way to success, if you’ve got something to prove, and/or if you just love data science itself, then chances are you’ve got passion and you’ll be willing to do whatever it takes to succeed. People like us are rare and I’ll be right there with you matching your passion as we show the world what we’re capable of together.