From academia to data science
Making any career change is challenging. Will my previous hard work count for anything? Do I have what it takes to succeed in a new field? Now, nearly three years into my career as a data scientist, I have been reflecting on taking the leap from academia to data science, a transition which was so daunting at the time.
Earlier this year I was invited to be a panellist for one of Europe’s top data science bootcamps, Science to Data Science. It specialises in taking PhDs in technical subjects and giving them the commercial experience needed to become successful data scientists. Back in 2017 I was an attendee at the very same bootcamp, filled with doubt and questions over making the change to industry.
So when they invited me back to join the panel, I was keen to share my thoughts. The other panellists and I reflected on our experiences working as data scientists, gave tips on technologies and careers, and answered questions from the audience on everything from what makes a good data science project, to how the role is likely to evolve in the future.
Many people considering a career in data science tend to worry about whether they have the exact résumé of skills required to be successful. No black belt in Spark and TensorFlow? What are you even doing here? This is an unfortunate consequence of technical roles like data scientist becoming defined by specific languages and techniques that are easy to list in a job advert. In my experience it is much more important to have an analytical mindset, well-versed in breaking down complex problems and finding creative solutions. Tools are just tools; you’ll pick up new ones in your career and put down old ones. What is key is knowing how to tackle a problem.
Multiple aspects of a PhD skillset are extremely valuable – likely your ability to visualise and articulate complex ideas are second-to-none. The formative PhD struggle of being left to figure out the best direction to head in an overwhelmingly large landscape is great experience for dealing with the ambiguity that can crop up in data science projects, particularly when defining the right problem to work on.
So why make the change? Data science is a tempting proposition if you have a penchant for technical problem solving, but want to see immediate, real, impact from your work. The variety and high turnover of interesting projects at a company like dunnhumby is a big plus compared to the slower pace of academic research.
While there are clear differences, it’s also interesting to reflect on the similarities between working as a PhD and a data scientist. One of my favourite things about academia was working in a small, international team of highly motivated, clever people. At dunnhumby, I still get to do that every day, with the added bonus of having a more diverse range of backgrounds to learn from. Physicists (my former tribe) are awesomely smart people, but in hindsight I see that we had all learnt the same things and tended to think in a similar way. Interacting with people from other fields like psychology, neuroscience, and computer science is a great learning experience and helps me understand many topics from a new perspective.
If you decide to go down this path yourself, I know you’ll also hit that wonderful moment of self-realisation: yes, my skills are valuable, and I really do contribute a lot to the success of a project. And who knows, maybe one day things will turn full circle and your esoteric PhD knowledge will come in handy in industry; quantum computers here we come!