Since I made the leap from academia to data science, I’ve become something of an evangelist for this career option — it’s one of the best decisions I’ve ever made! Over the past year, I’ve had many conversations with graduate students and postdocs who wanted to know more about how I made the career change. So let’s talk about it.
How did you make the decision to leave academia?
Let me start off by saying that I absolutely loved being a graduate student and a postdoc. I feel completely in my element when I’m figuring out how something works, so neuroscience represented an infinitely complicated puzzle for me to solve. Add to that my love of public speaking and teaching and it’s no surprise that I had always envisioned myself as a professor. But for about 5 years, I had been consumed with dread at the thought of moving to that next stage of my career. I was dismayed by the paucity of job opportunities and research funds. I wasn’t thrilled with the fact that I would have little control over where I lived and worked. I was concerned that an academic career would be too all-consuming and incompatible with raising a family while maintaining work/life balance. And I despaired at the idea of giving up a part of my job I loved (data analysis) to spend more time on a part of my job I hated (writing grants).
While postdoc-ing at UC Davis, I went to a meetup for women in data science and listened to a panel of speakers talk about their jobs. I was blown away by what I heard. These women had found fulfilling careers that allowed them to apply their research and analytics skills to solving problems outside of academia. It seemed like the solution to my career woes, but I required more data before I could make my decision. I scoured the Internet and read everything I could find about the day-to-day life of a data scientist. I reached out to my network and talked to friends-of-friends who had successfully made the switch. I sat with the decision for a few months because I wanted to be absolutely sure — I knew there would be no turning back. I got more and more excited about the possibilities that awaited me. In the meantime, I wrote a grant proposal (yep, still hated it).
Ultimately, I made the decision around June 2015 to leave my postdoc position at the end of the calendar year and pursue a data science career. This timeline allowed me to wrap up the project I had been working on, develop a toolbox of code for the lab, and apply to data science fellowships.
How did you prepare for a data science career?
While I was finishing up my postdoc, I tried to implement data science skills wherever I could. I switched from MATLAB to R. I started putting my code on GitHub. I developed a suite of functions that the lab could use for data pre-processing. I worked on side projects (see previous posts) to develop additional skills like web scraping and database querying.
I then completed the Insight Data Science Fellowship. This is an intense 7-week program in which you spend 3 weeks designing and creating a data science product (in my case, a podcast recommendation algorithm called thesauropod.us) and then spend 4 weeks demo-ing your product to companies and preparing for interviews. I highly recommend it. There are the obvious benefits: Insight prepares you to ace the Silicon Valley-style interview, introduces you to top companies, and provides you with an instant network of fellowship alumni who work at companies across the country. But on top of that, Insight gave me the confidence boost and the push I needed to actually go after my dream job. Had I been trying to prepare on my own, I think I would have felt like I wasn’t ready and would have limited myself to applying for the jobs I perceived to be attainable. I’m extraordinarily grateful to have been given the opportunity to participate and am so happy with where I ended up (Facebook was a mentor company in my session).
Do you have any resources you recommend?
Check out these blog posts by Trey Causey and Erin Shellman that discuss what it’s like to interview for data science positions. Read Drew Harry‘s take on selling yourself as an academic and read the responses to this quora question on the most valuable skills for a data scientist to learn. The hardest part of passing the data science interview is that you have to know a little (or a lot) about everything. Here’s what I found most helpful in preparing:
- SQL – Mode Analytics
- Python/R – Data Camp, Data Quest, this GitHub repository of python notebooks
- Probability/Math – Khan Academy
- Machine Learning Algorithms – sci-kit learn’s cheatsheet
- CS/SWE – Interview Cake
The tutorials are helpful, but they can only take you so far. Put your newfound skills to work and build something you love!