What PhDs do wrong (and right!) when applying for Data Science jobs

As a team with an opening for data scientists, we see lots of applications from people with quantitative PhDs. As someone who did a PhD myself, I’m really excited about what someone with this level of research experience can bring to our team. Getting a PhD in a quantitative hard or social science field requires tenacity, long term planning, a mix of broad technical skills, and clarity of communication about why your work matters to different audiences. Who wouldn’t want someone like that on their team?

Nearly half of people working as data scientists today have PhDs, though they make up less than 2% of people in the US over 25, so your academic training will command some attention:

“The inventor of LinkedIn’s People You May Know was an experimental physicist. A computational chemist on my decision sciences team had solved a 100-year-old problem on energy states of water. An oceanographer made major impacts on the way we identify fraud. Perhaps most surprising was the neurosurgeon who turned out to be a wizard at identifying rich underlying trends in the data.” (DJ Patil on “Building data science teams”)

Unfortunately, there’s a gap between the rhetoric pushing PhDs to industry jobs and what is necessary to actually get those jobs. I’m going to try to close that gap here by describing what great PhD applicants do, as well as some of the common mistakes we’ve seen.

The one thing that is non-negotiable for newly minted PhD students is write a cover letter!

With a short employment history that most definitely doesn’t include the title “data scientist” on it, the cover letter is your one chance to sell someone on why you could do a job you haven’t done. Be concise, but address the major questions someone might have about you:

  • Why does an oceanographer want to work on a team that studies how people use an app?
  • What attracts you about industry jobs versus an academic job?
  • Why is this team in particular exciting for you?
  • What skills and expertise do you have that’s not obvious from your resume?

Data science is an interdisciplinary field that requires frequent translation between engineers, product managers, marketers, and executives. The cover letter is your best opportunity to demonstrate your sophistication as a communicator.

If you make it to the phone screen, don’t presume you can talk your way through an interview in a domain in which you have little direct experience. You have to close the gap between the domains you’ve worked in historically and the domain you’re trying to move into by doing some serious background research.

At Twitch, the primary job of our data science team is supporting data-driven decision-making, especially around consumer-facing product design. Our primary collaborators are product managers. So we want to see candidates who have at least a basic understanding of topics like:

  • How is data collected about how tech products are used?
  • How do a/b tests work? What might make them hard to execute? What are their limitations?
  • How are ideas about “engagement” or “retention” operationalized?
  • Broadly, how do people interact with interfaces? How can interfaces fail?
  • What are the major dynamics in online communities?
  • How do organizations make decisions about products?

Every company has a different list, but they’re trying to tell you what that list is in the job description, so read closely! Figure out who you will be working with at a company and dig in on how those roles are constructed and how the team fits together. Look up the names of any major tools that you don’t recognize in a job description and try to work out what they do. Find newsletters or online communities where people talk about the role and follow up on any ideas that are surprising or new to you. Treat this process as seriously as you treated planning your dissertation topic and chase down anything that surprises you or doesn’t make sense.

The other major piece of the puzzle is understanding the company. You should spend at least half an hour playing with the product or reading about the company to understand how it works. What are the nouns and what are the verbs in the system? How do they fit together? How does the company make money? Who does it sell to? How does it grow? What makes it compelling to the people who love it? Not all companies will ask you detailed questions about their own product in a phone screen, but being specific about why you’re excited about the company or team will be valuable in nearly all situations.

Your most recent job was getting a PhD, which can be a bit of a mysterious process to many people. It’s important to reflect on what you actually learned in that process and how to explain it to people. There are two big pieces to this: selling your work and identifying your skills.

Assume your interviewer is not going to know much about your particular domain of research, and is definitely not going to be able to figure out why your contribution was compelling. You need to thread the needle between being too detailed (“we had four hypotheses, H1 was…”) and too self deprecating (“oh, it’s really obscure and boring”). Practice telling a story about what excited you about that work, what you learned, and challenges you faced. Non-academic friends can be a great resource here.

It’s also valuable to identify skills you picked up in grad school that might be relevant to the job you’re applying for. Did you excel in teaching? Did you manage your lab’s data collection infrastructure? Did you teach yourself Python to make a project go faster? Did you lead a team of researchers to accomplish something larger than you could have done on your own? Did you get a lot of experience writing and editing papers? Did you recruit grad students for your lab? These are all useful skills, so make sure your interviewer knows you have them.

Just because there’s clear consensus that people with quantitative PhDs can be great data scientists doesn’t mean getting those jobs is easy! Having a PhD is a reason to think you might enjoy that line of work and become successful at it. Don’t lose sight of the fact that having a PhD also makes you a high risk / high reward kind of candidate and take the steps you need to make it a smooth transition.

Thanks to Lian Chang, Danny Hernandez, and Brad Schumitsch for editing help. If you’d like to learn more about our team, you can start at science.twitch.tv or take a look at our job posting.

science @twitch, @medialab phd.