What not to put on your Data Science CV

11th February 2019


cringeFirst, I’d like to say this isn’t going to be a resume formatting session. What I’m going to talk about are some more data science specific issues; the things that will make any data science team reviewing your resume, cringe.

As always, there are no absolutes and there are always exceptions to the rule. However, as the person likely to be reviewing your resume these tips should do you well as a whole. Here are some things that could be on your resume that need to be addressed.

Don’t Name Every Model Under the Sun

There’s really no need to do this. It will only take-up valuable space. You want to allocate this to your hands-on experience. You don’t have to name every single type of regression, every single tree based model and all the ones in between. As a fellow data scientist, I can likely assume just because you only mentioned lasso regression on your resume doesn’t mean you’re completely oblivious to ridge regression.

I understand that there is often pressure to make it it through keyword based HR systems and non-technical folks, but there’s no need to go overboard. Your resume is supposed to showcase your experience. Not everything you’ve dived into a Wikipedia hole on.

Instead…

Choose the 5 to 10 you have the most comfort with and go with that. If you want some hints as to what models the company in question is looking for, see what they mention in the actual job description. Often times, you might just find a convenient list you can pull from!

Do not mention”toy” datasets as projects

I cannot emphasize this one enough. Don’t use the Titanic dataset for general ML. Don’t use Google’s 20 Newsgroups and say you have experience with NLP. Please please please do not say you have image recognition experience and then say it was just MNIST.  Never mention MNIST. It’s a huge red flag that you are not only a beginner scientist, but may lack creativity and originality.

These datasets usually just make me think you followed along with some generic tutorial. It doesn’t show you have the initiative and drive to attempt to solve your own data science problems, just that you can run the cells of someone else’s Jupyter Notebook. A solid portion of data science isn’t just modeling, but data munging, cleaning and troubleshooting. To apply these techniques on datasets outside the toy dataset space, you may find the need to employ new and creative approaches. Those are the skills I am looking to see.

Again, there are likely exceptions – maybe you’re a researcher piloting some novel model…or maybe you’re pre-training on generic open source datasets and applying a transfer learning approach on something else! That aside – there’s rarely a situation in which listing any one of these would impress me.

Instead…

Scrape your own data from the web and try to apply similar models you used on the toy dataset. Kaggle is okay, but what really impresses interviewers like me is original projects. In the real world, data science is about 1 thing – solving problems. Think about what problems you think you could solve with data science – what special expertise or domain knowledge you bring to the table. Showcase your value as an individual. That will be far more impressive than high accuracy on a problem that was already solved. Or was never even a problem.

Don’t Send the Same Resume to Every Position

While the interviewer won’t be able to tell if you tailored your resume for their company, it will hurt your chances as a whole not to. Early on, I fell prey to the mistake of bulk sending my resume to a ton of companies without revision. The problem with this is that it’s the lazy approach, and hurts your chances just a little bit with each company.

Instead…

Take the time to really read the job description. If you truly are qualified for the position, then you can tailor your resume to fit the skills on there pretty easily. This helps your chances of both bypassing their HR software, and getting the interview from someone who sees you have the skills that fit their specific needs.

Don’t be too General

I read a good article recently on medium about not being a data science generalist. The basic idea is to find your specific data science niche – there are many types of data scientists and getting stuck in the middle can stunt you. While personally I think everyone should get their feet wet in everything, as far as hiring is concerned this is great advice. A resume that is mediocre in all aspects of data science isn’t as good as a resume that excels in just one.

Instead…

Find your niche, and really portray what kind of data scientist you are. Don’t just list what you think companies want to hear, showcase what makes you different; the expertise you bring with you. Resumes often only make a brief impression, so make it memorable.

Read also: 7 questions you should ask your interviewer

Tagged with:

Kyle Gallatin

A data scientist with an MS in molecular & biology. He currently aids to deploy AI and technology solutions within Pfizer’s Artificial Intelligence Center of Excellence using Python and other computer science frameworks

To give you the best possible experience, this site uses cookies. Continuing to use jamieai.com means you agree on our use of cookies.
Accept