From Academia to Data Science

Data Science / Data Strategy / Careers

Being a great data scientist is mostly about the way in which you approach problems. Data scientists apply the scientific method in the commercial environment, which is why scientists can often successfully transition into the field. This sounds simple enough in theory, yet many scientists with superb academic backgrounds find it difficult to make the move into industry.

At ASI Data Science, we run Europe’s leading bootcamp for top quantitative PhDs and post-docs to transition from academia to industry. This September marks our sixth ASI Fellowship. We’ve worked with over 50 companies and helped over 80 Fellows find interesting and rewarding roles as data scientists and data engineers in industry.

What we’ve learnt is that companies have (often mistaken) assumptions about people who come from academia, and that academics often don’t help themselves by tailoring their language for industry. To counter this we built a program that tackled head-on the biggest reservations companies have of hiring academics. Here are the six main reservations, and how you can best prepare for them in your transition.

Can you work fast and complete projects?

A common concern companies have is that academics are more interested in ideas than products, never mind the dirty work of seeing things through to completion and implementation. The best data scientists don’t just complete projects, they are able to drive projects forward in a company and make sure their work helps change the business for the better.

Each ASI Fellowship culminates in Demo Day, where Fellows present the work they’ve done to a large audience of data scientists, companies and hiring managers. Eight weeks is not a long time to take a data science project from zero to completion so it’s incredibly important for Fellows to manage their time, the project and the company’s expectations effectively. Companies like to see a portfolio of completed projects rather than half-finished ideas, and it is much more impactful to have a smaller number of completed projects than a large number of unfinished ones.

Can you solve problems with dirty data?

Data scientists working in industry know that data is often unavailable, almost certainly imperfect, and project scopes and objectives are subject to change. The big dirty secret (or not so secret) fact of data science is that up to 80% of a data scientist’s time may be spent getting and fixing data. Good data scientists bring value to companies by being able to find projects with real business impact given the available data, and understanding when to pivot or change strategy. Problem solving under constraints is the bread and butter of an academic’s career, so make sure this comes across in your application and give concrete examples.

Do you have the key technical skills?

Data science is still a relatively new industry, and job descriptions often ask for experience in a wide range of trendy new tools and techniques. Regardless of the specifics of the role, we recommend budding data scientists invest in building a solid foundation in the core skills - Python (or R), basics of computer science (algorithms, code complexity, data structures), databases (start with SQL), statistics, linear algebra, and the basics of machine learning. There are many places to learn these skills for free or low cost online.

And once you’re confident you have the skills, make life easier for recruiters and companies by highlighting this on the first page of your CV, rather than buried somewhere at the bottom of page 4 after a long list of publications. Whilst the details of your particular research domain may be interesting (to you), for most roles what companies want to know is that you have the key skills that covers the bases. It’s no exaggeration to say that Fellows leave the Fellowship with entirely different CVs than when they come in.

Can you write commercial quality code?

Academics are rarely required to write reusable code, use a version control system or do code reviews. Learning to work in a team environment, collaborate on projects and read other people’s code is fundamental to data science work.

The best way to showcase your coding skills is through a portfolio, ideally on an industry standard version control system such as Git. Learn to be comfortable with version control, documentation, and keep your projects clean and readable. Having a genuine interest in open-source projects is always a bonus, and a great talking point at interviews.

Can you identify business value and stay focused on creating value?

For any project, the key is in understanding why and how the knowledge generated is important for business goals, not in generating knowledge for its own sake. Work produced by data scientists need to be usable by other teams to have impact, and often this means managing expectations well. Producing great work but answering the wrong questions also has very little, if any value. What’s interesting may not be what’s important. Academics have a reputation for looking inwards, rather than how their work relates to what others are doing. In changing environments when projects are not well defined, it’s especially important you understand what is expected of you, and align your work with the goals of the company.

It’s good practice to have business value in mind from the very beginning of a project. A good project proposal includes a comprehensive background on the project, defined objectives and potential impact, how a successful project may be implemented, the key stakeholders, as well as the datasets available. Practice applying this methodology on the projects you work on and try to find a quantifiable way to measure impact wherever possible. On the Fellowship we also run a number of interactive workshops to help Fellows see beyond the technical sides of the project.

Can you communicate at the right level?

And lastly, this is the biggest fear of companies we speak to. Very few data scientists get to spend their days only communicating with other data scientists. In most companies, the work data scientists do will impact the product team, marketing team, sales team and many others, and being able to build relationships with non-technical colleagues is key to getting your work implemented.

Although in academia there are various opportunities to communicate with others (such as at a conference, in a seminar or as a teacher), you rarely receive feedback on how you communicate and whether it is effective. Without feedback and deliberate practice, it is difficult to get better. During the Fellowship we put a lot of emphasis on communication and presentation skills. Fellows take part in a series of workshops designed to take them out of their comfort zones and improve their skills from many angles. If you want to push yourself, try entering your local IGNITE event.

Changing career is a big step and there’s a lot to cover. We suggest treating it like a project, break down the steps systematically and make sure you’re not making unnecessary mistakes. At ASI, our method is the Fellowship, it’s free for Fellows to attend and we run them three times a year. Find out more and apply here - (http://asidatascience.com/fellowship)

Share on: Twitter, Facebook, LinkedIn or Google+