As a data science instructor, I get asked about my path to data science quite a bit. Diversity of expertise is a strength of our field. At the same time, this lack of a defined pathway makes it daunting for an outsider to figure out how to jump into the swirling whirlpool that is data science.
When we interview potential bootcamp applicants at Metis, we look at their statistical background and coding background, but then we add a personal interview. Part of that assessment is an understanding of four key qualities:
That emphasis is part of what drew me to Metis - a good data scientist isn't just someone who can code and rote repeat some algorithms. It's someone who is inherently curious about the world, able to attack a problem from different perspectives, wants to answer the hard-to-answer questions, and is okay with failure.
These skills are also vital to becoming a data scientist. In this series, I'll talk about the non-programming, non-stats skills vital to marketing yourself as a data scientist.
I'm not going to laundry list coding and programming and modeling skills - there are a ton of resources online that enumerate those for you. The point of my blog posts are to show you how to leverage those skills into a job.
From Academic to Data Scientist
I could cite a list of schools and degrees, but I won't. My interest in data science boils down to something really simple - I have always been interested in explaining human behavior using data. It's the best combination of creativity and objectivity.
At some point during my PhD program, I realized that academia was not for me. The budding field of data science was the perfect mix of answering big questions using data, but also creating real results with it. People would take what I did, and take action on it.
How did I re-invent myself as a data scientist?
First, I read all I could about what the industry and what it meant. There's the obvious stuff, like taking the free online courses in Coursera, Udacity, etc. What a lot of my research told me was that academic types are often hard to hire because they can't work in 'real world' settings. I had a history of working in analytics and forecasting, and I knew that the classroom setting and the real world were two totally different animals. I had to prove that I not only had the hard skills, but I could execute a project as a team leader and member.
To do that, I leveraged my network. My masters advisor was a valuable mentor, and he was well connected in the non-profit sector. I also joined Gary King's Gov 2001 class, which is a class on replicating and advancing journal-published papers...that also has a stellar alumni group and Facebook page. I was already part of my masters program alumni group mailing list, and I made sure to find other mailing lists in my field, no matter how remotely related. Yes, that meant tons of spam, but It led to two or three opportunities I would not have otherwise had, including a position at the World Bank.
If you're interested in learning more about my projects, check out my resume or my linked in. My first project was a lower level stats role - I worked as the analyst on a team to solve a problem. After I had a few under my belt, I launched out on my own, and provided end-to-end solutions. For one of my projects, I did qualitative interviews by phone, designed a survey in Qualtrics, and ran the analysis afterwards.
My strategy was to take jobs that showed different skill sets: model complexity, programming skills, the ability to work on a team, the ability to conceptualize and implement a project, and the ability to communicate results effectively.
I was selective in what I spent my time on. Consulting is a huge time sink, so I made sure my projects reflected a wide variety of skills implemented in different ways across different industries. One bonus was that I added a handful of excellent references that were relevant to the industry.
Finally, I built out my resume and contacted recruiters. I used Burtch Works, and had an incredibly helpful conversation with Linda Burtch. She advised me on some resume tips, as I had an apparent gap in my work because of graduate school, and she gave me some basics on what my expectation should be for salary. She also told me the most valuable piece of information, which still holds true - it's incredibly difficult to break into Silicon Valley, but once you're in, you'll not have a problem getting another job in data science. Ultimately, the position I received was not through Burtch, but I keep in contact with them, and make sure to fill out their annual salary survey.
Later posts in this series will go over specific skills and tasks and how you can use your resources to get hired as a data scientist.
Tl;dr - In transitioning from academia to data science, I realized I couldn't just call myself a data scientist and assume jobs would come my way. I took on consulting positions in a variety of industries that showed different skills to prove to potential employers that I could handle the variety of non-programming, non-statistical skills needed to be a data scientist.