FIU Data Science Bootcamp Application Deadline is Oct. 3 - Apply Now

Navigating the Data Science Job Market - What to Know Before You Apply

By Andrew Savage • January 24, 2018

This post was written by Andrew Savage, Metis Sr. Career Advisor and Head of Employer Partnerships. 

Photo by Andrew Neel on Unsplash

Within the last few months, I’ve given talks at ODSC West and the Global AI Conference where I shared information about the data science job market. Because of the positive reception to both talks, I wanted to share these perspectives more widely here on the blog. My goal is to help anyone looking to break into the world of data science as a job applicant.

For the last 2 years, I’ve been managing career placement and employer partnerships for Metis from our office in San Francisco. During that time, I’ve helped hundreds of our alumni get jobs as Data Scientists, Machine Learning Engineers, and increasingly, AI Engineers.

What’s great about this experience is that I’ve seen just about every type of person come through our 12-week program and successfully transition into data science. From ages 19 to 60, from those fresh out of their undergrad program with no work experience, to those with decades of professional experience. From Master's and Ph.D.’s in Statistics and Computer Science to Associate’s Degrees in Photography and everything in between.

So What Does a Data Scientist Do?

This is perhaps the toughest question to answer, because as you’ll come to find out, data science is less a specific role, and more a wide variety of processes applied to a wide variety of problems. Not every role requires the use of every data science process.

To make an analogy: as you become a Data Scientist, you’ll have a tool belt. And as you learn more and more techniques with different types of data, you’ll be adding tools to your belt. Sometimes you’ll find yourself feeling fully proficient with a number of tools, while other times, with just a few.

In following this analogy, we also arrive at an important lesson: Just because you can wield a hammer, doesn’t mean every problem includes a nail. Similarly, just because you were able to solve a data problem using (insert model and methodology of choice) doesn’t mean it’s appropriate to use the exact same methodology for another data problem.

For instance, one of the things we teach at Metis is how to build recommendation systems. These are tremendously useful machine learning models that can be used for a variety of problem sets – most notably in the field of media and entertainment (think Netflix) or online retail (think Amazon). But not every data science role you find online and apply to will require you to build a recommendation system. It all depends on the company’s goals and the current problems the data team needs to solve.

You’ll need to begin thinking about these things as you apply to jobs, which leads me to the application process...

How to Stand Out in the Application Process

People often think that applications are simply vetted by way of merit, i.e. the person with the better degree, more work experience, and better portfolio will get selected.

This is only partly true, and even at that, only true IF the recruiter or hiring manager has time to actually LOOK at your application.

Let me demonstrate with a visual.

I took this screenshot about 24 hrs after the post was created. Look at how many people responded. 848! Now you may be thinking “Ok, but it’s way easier to type “Interested” than it is to apply for a job.” But is it?

For many open jobs, a “Quick Apply” option is now available, meaning that one-click job applications are a new reality. And even if that option isn’t available, a standard resume and cover letter attachment can be done in about 1 minute.

There are pros and cons to jobs being posted online and discoverable by anyone. The pro is that everyone can find and apply to jobs quickly and easily...and the con is...everyone can find and apply to jobs quickly and easily.

As a recruiter or hiring manager, there is NO WAY to properly scan through 500+ applications in any sensible way. In many cases, maybe a few dozen get chosen at random and looked at seriously and things are paired down from there.

As I often tell my students, you’re not simply vying for the position against others based on merit; you’re vying for your application to be SEEN and READ. If you can achieve those goals, you’re in elite company already.

How do you make it happen? The key to the application process is differentiation. Primarily, we’re talking about differentiation through your application METHOD.

So if you know most people choose the path of least resistance (the quick online application portal), you need to go through a backdoor method. That is to say, find a way to relay your message to a person who will be responsible for pushing applications forward, but do it in a way that most others wouldn’t think of.

Here are some suggestions: Find a recruiter and/or hiring manager through LinkedIn; send them a personal message expressing your interest. Or, guess their email address or tweet at them. Obviously, the best method, above all, is to use a personal connection at the given company or to get a referral to someone there, but if that’s not an option, these are fantastic ways to stand out.

When it comes to what to say in your message, don’t overthink it. You should limit the message to 250 words and your goal is simple: relay that you understand the problems they’re working on by putting forward an example of something you’ve worked on in the past that shows you’re capable of solving a similar problem.

This is ALWAYS a winning formula and will make you stand out, even against those with advanced degrees and lots of work experience.

The Key to the Interview Process: Relevance

Once you get selected and invited to an onsite interview, your chief goal is to showcase how you are the solution to the team’s problems.

This is an EXTREMELY useful skill in any interview but also specifically in Data Science. As you’ll come to find, every company and team will be working on a slightly different problem and it will be YOUR job to position yourself as the solution.

As we’ve already covered, data science is less a role than it is a process to solve a wide array of problems. So, the goal of your interview is to continually uncover the nature of a company’s problem, understand their process for solving it, and discuss any relevant work that you’ve done. This will convince them that you not only understand what they’re saying but that you have the relevant experience and knowledge to help solve the problem.

To go back to the tool belt analogy for a moment – don’t go into the interview assuming that all problems have nails, because if you do, you’ll only be mentally prepared to showcase your hammering skills. And while those skills might be valued for some roles, there are other roles that will need your saw, wrench, or whatever else will get the job done.

So what’s the right project or experience for you to highlight? In many cases, it’s safe to lean on one that showcases relevant domain expertise (ex: maybe you built a travel app and you are now interviewing with Expedia). Or even better, a project that showcases your familiarity with solving the team’s TYPE of problem, regardless of which subject domain you previously covered. For example, say you did a data project on the electrical grid where you looked to optimize pricing based on variable energy conditions. Your first thought might be that project might only be attractive to a PG&E or a SunEdison or some other major energy utility. But what’s the nature of that problem? Isn’t it essentially a supply/demand forecasting problem? Think of how many companies have problems related to that sort of situation. Uber and its Surge Pricing? AdTech with its real-time marketplaces? Airbnb with its user/host forecasting? That’s what I mean by showcasing a relevant type of problem. These will help you win over your interviewer, even if the industry use-case isn’t related.

To recap, I hope you remember that data science is not one giant conglomerate of roles. Every one of them is going to be slightly different. In fact, I urge you to beware of following jobs based on title alone. I know many Product Analysts working on cool machine learning projects, as well as Data Scientists mostly building dashboarding tools for basic analytics reporting.

Hone in on what a company is really doing and trying to achieve through the data they collect. From there, be creative with your communication methods and your way of grabbing someone’s attention. And once you get the interview, highlight the relevant skills and experiences you have that will help your employer solve their problems.


Learn more about the Data Science Bootcamp and download the 12-week Careers Curriculum here

Similar Posts

7 Keys to Success in Your New Data Science Job (by Metis Sr. Data Scientist Jonathan Balaban)

By Jonathan Balaban • October 10, 2018

Over the past decade, Metis Sr. Data Scientist Jonathan Balaban has helped more than 100 students, colleagues, and friends start new careers and find fulfilling jobs. In this post, he provides 7 simple but powerful keys that can help you make a great first impression and accelerate your professional development.

5 Reasons Why You Should Move to Chicago for a Data Science Job

By Ashley Purdy • October 30, 2018

What makes Chicago a great place to launch your data science career? And how does it compare to other big cities like SF and NYC? Here are 5 reasons why you should consider a move to The Windy City.

Love Math? Love Stats? Here’s How a Data Science Bootcamp Can Help You Find Career Fulfillment

By Emily Wilson • June 27, 2019

There are a number of individuals with careers in banking, accounting, education, academia, and other fields who are seeking professional change for a variety of reasons. How can a data science bootcamp factor into that scenario, especially for those with a passion for math and statistics? Find out here.