Leaving the academic canyon

I’m leaving my career in academia as an evolutionary biologist to take a position as a data scientist. Yes, the hype is true: businesses do want people with analytical and computational skills. I’m excited about this move because it allows me to continue applying my analytical skills even bigger data, and learn new skills along the way (hello Hadoop!). Equally importantly, it allows me to spend more time with my family in a place we love.

Many people have written about leaving academia [1], so here’s my contribution. Unlike others, my story is mostly happy, maybe cautionary.

Looking back on how I got to where I am, I feel the best analogy is going for a hike in a box canyon. At the start, the canyon is wide, beautiful and seemingly endless. About half-way down it starts to get narrower, but you don’t worry because it’s still beautiful and you’re enjoying yourself. But then, you get to the end, and the only way out is a steep climb to the top.

correlation plot Box canyon near Sedona, AZ. Photo: John Stanton-Geddes

I got to the end of the canyon, and I could see the path out. I even think I could have climbed there. But I’d had a good long hike, and surprisingly, I found a side canyon. Climbing out was no longer the only, or even the best, way to continue.

Back to reality, when I started my PhD in 2006, I had no skills to speak off. I liked to teach. I liked biology. Getting a PhD sounded fun (and it was!) and what other choices did I have with my liberal arts degree? So off I went to get a PhD at the University of Minnesota. They were a tremendous 5 years. I learned enough to become a quasi-expert in my sub-sub-field, got to do some great field work and lab work, taught undergrads, made great friends, and got married. My PhD advisors were the best you could hope for and only have (and still do) provide me with encouragement. My postdoc mentors have also been great and supportive. I like to think that I had a promising academic career. Hell, I’m up to 82 citations 3 years after defending my dissertation (Google Scholar Oct 8, 2014) and in the process of submitting a great paper to PNAS (where it probably won’t be accepted, but it’s still a good paper!).

The catch is I now have skills of value. I wish I could remember what blog or twitter post I saw this on, but it turns out that many of the characteristics that make a successful researcher are the same characteristics that make someone valuable to industry. I picked up a minor in statistics and am a reasonably confident statistician. I spent a lot of time working in R, and actually found that I enjoy programming. Which is ironic given my main memory of ‘Intro to Comp Sci’ in college is that it was the first (and only) class I skipped on a regular basis. I started to learn to program in other languages, how to use linux, how to work on a server, and other skills that are generally associated with the term data scientist.

Thinking towards the future, here’s what pursuing a career in academy would likely require:

Apply to 20 (or 40 or 60!) academic positions across the country.
If lucky, get asked to do an on-campus interview at at least one (maybe two!) institutions.
Light candles and pray that the stars align so I get offered the position, with (1) decent salary (non-negotiable as set by university or union policy), (2) reasonable start-up so I can do research, and (3) institutional support to succeed in teaching. For what it’s worth, the last would have been the most important and probably least likely of my requirements.
Teach 1-3 classes per semester consisting of a mixture of motivated and un-motivated students paying the price of a new Tesla each year.
Spend hours writing brilliant grant proposals with about a 10% funding rate.
Work my ass off so I get tenure or can “trade-up” to a better institution or place closer to where I want to live.

Context: my postdoc funding runs out at the end of 2015, so I kinda sorta need to get a job this academic hiring cycle. I have two kids so I need a job. We live close to my wife’s family so the incentive to move is low.

To paraphrase something I read somewhere I can’t remember: “If I treated my wife the way science treats me, she’d have left me long ago”.

In contrast, the data scientist position took three interviews and I was offered the job about 6 weeks after hearing about it. Salary was better than my median expectation as a starting professor, I asked (and got) more vacation time. Of course, long-term this position brings up new challenges such as will I succeed in the business setting, will my company value data scientists, and what are my long-term goals. None of these challenges are insurmountable or greater than the academic ones listed above. They’re just different.

So, it turns out that the continental shift from academia to industry was actually quick and easy.

Another related issue is that the people I respect and look to as examples changed. As a student, it was my professors, and for the most part, I still have tremendous respect for them. But the more time I spend in the analysis world, I’ve found role models such as Hadley Wickham, developer many great R packages who left his academic job to work at RStudio, Yihui Xi, also now at RStudio, and Hilary Parker, data scientist at Etsy, that are doing exciting work outside of academia. They set a great model for success, and in a way that directly contributes to their communities (i.e. tax dollars) to pay for universities and NSF grants.

I don’t think there’s a clear lesson here. I’m just another data point in the figure that less than 10% of PhDs become tenure-track faculty. I don’t regret any of my decisions. I’d never heard of R before I started my PhD and certainly couldn’t tell you what a PCA was. I learned those skills during my PhD, and had a great time doing so. It may have taken 3 times longer than if I’d just gotten a masters, but it also didn’t cost my anything other than my time. I got to meet many great people, think about important questions, and contribute to valuable research. In the end, it turns out that it is hard to have it all, for men as well as women.

So long, and thanks for all the data.

[1] I’d fall into the ‘Explainer’ category, which is consistent with my philosophy to make my scientific work as open as possible…including leaving.