I have accompanied quite a few PhD students and postdocs in the transition to a job in industry. Even at a later stage – that is a PhD of 5 years and more – this career change is typically manageable in a few months. Demand for new talent in the fields of data science and artificial intelligence (AI) is increasing, particularly also for highly numerate PhD students with experience in handling numbers, images, text, and speech; and prior experience in using Python, R, and/or SQL.
When interviewing candidates for a transition to data science, my top three concerns are:
- A background check for cognitive ability and experience in handling data,
- The motivation for a move to data science or AI, and
- A sense of direction or destination in making that move. Which field or industry do you want to join?
In managing your transition, prior coding experience in Python or R is very helpful but not essential, as it can be acquired rapidly with a combination of books, online learning and immersive coding experiences.
In empowering PhD students and postdocs to move to data science and AI, I would like to do the following:
- Share some results from an information campaign that I have been rolling out for PhD students and postdocs across Germany.
- Discuss some pros and cons of continuing in academia or moving to industry as they have emerged from the workshops I’ve held to date.
- Address the gap between science and data science and how you might close the gap so that a hiring manager will see you as ‘production-ready’, e.g. able to improve and deploy predictive models.
- Look at how drafting an industry-relevant CV might accelerate your transition, which, once you start, should take you 6 to 9 months.
10,000 Data Scientists for Europe is a PhD information campaign that I have taken to Göttingen, Bonn, Köln, Tübingen, Stuttgart, Heidelberg, Freiburg, and Karlsruhe, as well as running multiple smaller workshops in Berlin. The 12 workshops have had 576 registered participants and I have collected feedback from 236 attendees.
- The largest workshops were Heidelberg (109), Bonn (98), as well Tübingen and Göttingen (76).
- More than 60% of the respondents think a career change to data science very likely, and more than a quarter are definite about the change.
- More than 60% want to transition within 12 months or less.
- More than 80% have prior coding experience with Python or R.
- 4 out of 10 can imagine being a co-founder of an AI-driven startup.
At the workshop, I ask participants to discuss the meaning of continuing along the academic track or changing career by moving to industry. Of course, it makes a difference whether you are doing a PhD, already are a postdoc, or have prior industry experience. Yet, some salient points have emerged.
While the low pay, temporary contracts, and statistical odds of achieving a sustainable academic career are a great turn-off, the ‘academic dream’ is alive with many who emphasize the freedom, passion, and personal satisfaction that comes with research. From the workshop, I have data suggesting that most earn less than €40,000 per year in academia, while I consider it reasonable to ask for €60,000 or more as a data scientist with a PhD.
A more critical take is that the structure of academia demands a narrowing specialization that increases personal risk (e.g. of later unemployment) while a move to industry enables the building up of a broader portfolio of projects, methods, and technical skills. I support individuals in this transition and see that significant career progress has been achieved after two to three years, e.g. moving on to more senior roles.
Most PhD students and postdocs associate a move to industry – quite generically – with more security, stability, and income. Interestingly, many also expect a broadening of opportunities by achieving real-world impact through data science product development. And yes, often there is an interface to e.g. customer experience, business revenue, research and development.
As the workshop facilitator, I can offer additional observations that address the difference between academia and industry:
- A significant difference is that early-career researchers often work individually with longer time lines to publish their results, whereas industry teams work in shorter cycles, possibly with daily deployment routines.
- Perhaps the biggest challenge when leaving academia after many years is to get one’s head around the notion of the use of business cases. Not peer recognition but rather user feedback and paying customers matter.
- Industry and startups expect a production-ready data scientist, so setting aside some time for a data science project with a demonstrated and reproducible outcome is valuable for landing the desired first position.
To empower PhD students and postdocs in their move to data science and close any gaps, I have compiled a simple roadmap. I suggest organizing the transition in four steps:
- Exploration of the field: First, get a feel for and understanding of the role of the data scientist by utilizing e.g. courses, hackathons, meetups, and by interviewing practicing data scientists and credible recruiters. For you, this should result in a stop/go decision.
- Domain orientation: Which field or industries are you considering? Are you more interested in computer vision, big data or natural language understanding? For health, finance, or automotive industries? As you consider your options, look for indicators as to how high or low the entry barrier is. For example: How new is the product? How large is the industry? Are startups hiring aggressively and must the wider industry follow suit (e.g. autonomous driving)?
- Further training (if any): Your interactions will give you an idea if further training is required for a successful transition. If in doubt, you can also interact with training providers (e.g. data science boot camps) to see what gets their graduates hired in the domain you are interested in.
- Career entry: I reckon you want the learning curve to be steep, so a team with a good track record in a vibrant urban environment may be the first choice. A good track record is indicated by a growing team, team members staying for at least two years or more, a product on the market, and growing revenues.
There is a way that you can make your move to data science much more focused: By working with an industry-relevant CV from the start. This means writing a new, second CV that you take with you for your interactions and conversations to collect feedback. Your interlocutors and respondents may much more easily have some of the following for you:
- Suggestions of which domains might be interesting and accessible for you.
- Network contacts that may be interested in your CV.
- A good idea of what the gap (if any) is vis-à-vis your preferred job, and how to close it efficiently.
What do I mean by industry-relevant CV? A presentation tailored to hiring managers and recruiters or human resources, making it clear just how you and your skills are relevant. Such a document is always individual. Still, I can offer the following guidance:
- The first page should include your mission and search statement, a technical skills overview, and your last employment.
- The second page includes your education, any other employment, and the transferable skills you bring from academia to industry.
Let’s look at this in some more detail. Your mission and search statement says who you are (e.g. Data Scientist), what you want to do, where you are looking, and why you want to be a Data Scientist. Next, you should list all technical skills that you are confident of and ready to be tested on. I believe it helps if you indicate your confidence on a simple scale, e.g. a 3-point scale. I recommend treating your Ph.D. or postdoc position as employment and making the effort to describe it in ways that would be relevant to a hiring manager or human resources department in industry.
When listing your achievements in education you may include any data projects or relevant student jobs you performed. This is to say, I do not recommend listing teaching or student jobs in the employment section, particularly if they were short-term. However, looking back at your activities and achievements over the past years: Which highly transferable skills have you acquired? Please do ask a search engine for help, and decide which three to five transferable skills you want to list and provide two specific examples for each skill.
Good luck with your transition to data science and AI. Not only do we need more and better talent in this field, but these are also exciting times and the dynamic development of new companies, products and services will carry us for many years to come. If you would like to keep up-to-date with the campaign and its workshops, you can do so on Eventbrite, Facebook, Medium, or Twitter.
Dr. Chris Armbruster is Director, Data Science Retreat, Berlin. In early 2018, he launched the campaign 10,000 Data Scientists for Europe with the aim of finding and empowering within five years 10,000 talents for AI-driven product development.