Alumni Spotlight: Phillip Schafer Talks Transitioning to Industry and Advice for Applicants

I got started doing computational physics as an undergrad at Yale. I had a summer job in the Geology Department doing numerical convection simulations and found that I really liked writing code to solve a problem. After college, I taught high school physics for a year and then went back to the Yale Geology Department to work on data analysis for a precipitation study. Back then my toolset was limited (Excel, Matlab, a bit of C) but I got to cut my teeth on finding interesting patterns in data.

When I started grad school at Penn State, I thought I wanted to be a more straight-ahead experimental physicist. I worked for a while in an atomic physics lab, but found that I missed writing code. I looked around the physics department and found a group that was doing computational neuroscience. My advisor, Dezhe Jin, had an idea for a project using ideas from neuroscience to design better speech recognition systems. I enjoyed building up a new way of attacking an old problem, more or less disregarding the standard methods used in the field. I also got to try my hand at a lot of machine learning and statistical methods in the process.

As I was finishing my PhD, I was looking around for new and interesting problems to work on. An email about The Data Incubator circulated around the physics department and I thought I’d give it a try. I knew I’d made the right decision because I had a lot of fun doing the take-home problems in the application. I had to apply twice before I was accepted, but my persistence paid off.

The Data Incubator was a fantastic introduction to the world of data science in industry. Curriculum-wise, the course was amazingly comprehensive in covering the tools and techniques most in-demand in industry today. Many of the topics we covered came up in interviews and problem sets during my job applications, and I wouldn’t have done nearly as well as I did in my job search without the skills I gained in the course.

Another great thing about the program was the opportunity to hear directly from company representatives during the panel discussions. This helped me get a feel for the lay of the land and discover what types of jobs were out there. There were a lot of companies where I never would have considered applying, but once I heard what they were doing with data and about the scope of their technical ambitions, I became really interested.

Learn as much as you can about data science in general, and in particular about the Python stack (Numpy/Scipy, Pandas, Scikit-learn) before applying. The first time I applied, I was able to solve the take-home problems but my code was sloppy because I didn’t know the proper tools. I took some time to learn more Python and wrote much more elegant code the second time around, which seemed to make the difference for me. That investment of time will also make your life a lot easier once you start the Data Incubator program. The workload is intense, so the more you know going in, the more time you’ll have to actually focus on getting a job.

As an Online Fellow, I did a series of weekly assigned mini-projects during the course. The topics ranged from web-scraping and database projects early on, to machine learning and NLP in the middle of the course, to “big-data” techniques using Hadoop and Spark at the end. The projects were a lot of work but they really helped cement the course material for me. The single large capstone project (done by the NYC Fellows in our cohort) has its own advantages, but I was ultimately glad to do the mini-projects because they allowed me to digest the course material more thoroughly.

The beginning of the program can be a bit overwhelming at times, because there’s so much coursework to do but you’re also supposed to be reaching out to companies where you’d like to apply. I found that most companies were happy to give me a few weeks to focus on the mini-projects before starting the interview process. Those early reach-outs were important, though – I first contacted Optoro by “liking” them on The Data Incubator’s system during the first week.

On a related note, be sure to get your resume in good shape early on, and ask for feedback on it from The Data Incubator’s staff. At a lot of companies, your info gets passed to someone who’s not familiar with The Data Incubator, and so all they know about you is what’s on that one piece of paper. I ended up removing a lot of fluff about awards and honors from my resume, and added more specifics about the skills I picked up through The Data Incubator and my PhD.

The best general job-hunting advice I heard came from Jon at The Data Incubator. He compared the application process to a funnel: you have to put lots of stuff in at the top of the funnel in order to get just a few offers to trickle out the bottom. The worst thing you can do is to attach your hopes to a single job and to stop feeding the funnel. Being in The Data Incubator doesn’t guarantee that you’ll get a job, but it does give you a leg up over the competition. Keep an open mind about the available opportunities, and be confident in the skills you’ve obtained. Good luck!