Getting hired as a data scientist with a STEM background was much harder than I initially thought. I had naively assumed that a PhD in physics (and triple-major in astronomy, math, and physics) was sufficient qualification for the job and, technically, it was but I struggled quite a bit to find employment. Despite the red-hot job market, I applied for upwards of 100 jobs, wrote roughly 30 cover letters, and received only a handful of interviews over a six-month time frame. In hindsight, however, I realize now the simple changes I could have made to drastically strengthen my resume.
The problem with my early applications was that I severely underestimated the extent to which modern data science is virtually synonymous with machine learning. I knew that I would likely be doing machine learning on the job but I didn’t know how important it was to demonstrate competency with machine learning before applying for a job in data science.
In each of my four interviews, I was asked a technical question about machine learning that I simply couldn’t know the answer to with only a passing familiarity with machine learning. Indeed, one job didn’t even interview me but instead sent a data set and wanted me to compare and contrast different machine learning models. I had assumed that employers would take it for granted that I would learn on the job, but the level of detail required to excel at this step of the application process implied that they were really only looking for candidates with prior experience building machine learning algorithms.
To address this, I took Andrew Ng’s Machine Learning Specialization on Coursera. This specialization broke the theoretical underpinnings of Machine Learning down into five courses, each of which could be completed in about 10–20 hours of work. For anyone trying to get a job in data science, I simply can’t recommend these courses strongly enough. By the end of the first course, I felt a monumental shift in my relationship with machine learning. Simply knowing the fundamentals of machine learning allowed me to properly engage with interview questions I was posed with, rather than sidestepping them with generalities. In addition, the courses provide official certificates which integrate with your LinkedIn profile and go a long way in reassuring potential employers that are capable of performing machine learning tasks.
Sadly, I’m fairly certain that it was the $60 machine learning certificate, rather than the 10 years of tuition, that got me a job, as immediately after catering my resume to highlight machine learning, the frequency at which I received interviews went from 4% to 100%. The next three jobs I applied to all called for an interview and I accepted a Senior Data Scientist position offered by the first one.
In conclusion, the biggest takeaway from my prolonged job search is that you need concrete machine learning experience to get hired as a data scientist. Fortunately, this qualification is relatively easy to obtain if you are coming from a STEM background. Online courses offer extremely high-quality content that can be completed in a reasonable amount of time and offer certificates that reassure potential employers. Better yet, this small amount of training is more than enough of a background to succeed in the role I now have, which relies primarily on off-the-shelf machine learning algorithms.
