• Data Science
  • Ben Taylor
  • MAR 05, 2018

Getting That Data Science Job

Ready to learn Data Science? Browse courses like Data Science Training and Certification developed by industry thought leaders and Experfy in Harvard Innovation Lab.


like a baby bird jumping from the nest

Most people who attempt to get hired as a data scientist fail. This article is to help clarify what is happening and increase the chances of landing your first data science job.

I have been on the hiring side for data scientists and I have seen others in industry struggle to get offers while their counterparts get every offer. Here is my distilled noodling on the topic:

Rising Above The Masses


Ok, I'm just going to rip this band-aid off: "You are not interesting right now"

Everyone looks the same, literally. How do you expect a hiring manager to put you in their top 5-10 list for the final round if you look just like everyone else in the applicant pool? Everyone has a basic machine learning background. Everyone has the same skill sets listed (i.e. python, sklearn, xgboost, pandas, numpy, etc...). Everyone has the same projects (MNIST, sentiment, etc..). Everyone has the same educational background (Northwestern, etc..). You have to keep in mind that the good data science jobs that you want probably have 100-1000+ applicants. Most applicants also lack any meaningful work history, they are all trying to break into their first data science job.

Once you admit that your fancy resume is actually boring and you look just like everyone else (believe me you do) you can start making some meaningful changes. Tip: Add something that people don't have. Create a new useful machine learning library/contribution and share the GitHub link. Have a data blog? Have a data youtube channel? Are you are code contributor to Sklearn/Keras/Etc.. Did you write a book? Do you have a video link of you presenting at a local meetup? Have you done some consulting? Did you make Tensorflow suck less? Were you the first person to demonstrate GPUDirect in mxnet (nobody has done this yet BTW)? Do you have a machine learning patent/publication? Are you top 1000 globally on Kaggle? Are you special?



For a bunch of introverts, this is hard. So if you know you will be one of many identical applicants in the pool one way to improve your chances is to network. Get involved with local meetups, get to know people, and make sure they know you. I always recommend presenting at a local meetup on something that intimidates you. Good networking can help guarantee you are in the top 5 consideration if you can get a local data science leader to vouch for you. I have seen people go directly to the #1 ranking candidate based on a personal referral. Tip: My main tip here is to start presenting at local meetups, broadcast yourself.

Know What Industry Wants:


Most applicants are clueless about what industry actually wants/needs. Unfortunately, many formal data science programs aren't helping here either. There is a huge gap between what academia teaches and what industry wants. An example that comes to my mind is the reverse skew of R being taught vs. python in academia and there being more python data science jobs than R data science jobs. Another issue is most institutions can't afford the latest talent on hot topics such as deep learning, etc... therefore you are on your own. In the end, this is the candidate's responsibility to know what is needed and if there are gaps there are plenty of books and online resources to fill them. Tip: Networking and following industry leaders will help you know what gaps you may have between what you know and what they need. You may even just ask some hiring managers what the top 5-10 skills are that they are looking for in an applicant today.

Humans Aren't Good At Hiring (Bias + Luck):

I am not sure if there is any comfort or value in this realization, but humans are still terrible at hiring in 2017. We all have an unconscious bias based on our backgrounds and experiences. We are more likely to hire based on similarity or likeability than KPI related competencies. There is also a luck component where you may never hear back from a company because you applied too late in the application cycle, no fault of your own. All the more reason to avoid the main hiring funnel and make those connections more personal and help the humans out. Make yourself such a compelling candidate that you force people to hire you, they would be idiots not to.

Does Education Matter?

Only for boring companies, you don't want to work for them anyway. What you can demonstrate and know matters more than what certification or educational title you have. I have hired PhDs data scientists and I have hired college dropout data scientists. Some do better with a formal education while others do better with their own hacked curriculum, figure out what you need. If I had to do it over again I would skip college completely, 0, zip, nada.

The Harvard Innovation Lab

Made in Boston @

The Harvard Innovation Lab


Matching Providers

Matching providers 2
comments powered by Disqus.