Career Chat: Susheela Singh, Senior Data Scientist at Google
A few months ago, I was a volunteer mentor for the Data Scientists in Training (DST) program run by Dr. Eric Chi at Rice University. The DST is an outreach program designed to introduce high school students to concepts and careers in data science. As part of the program, a number of people came to talk about their careers in data science with the program participants. Dr. Susheela Singh, a data scientist at Google, was one of these people and she gave me permission to share that discussion with you here. This post contains some highlights from things she shared during her talk. Thanks so much, Susheela!
Susheela grew up in Houston, attended college at the University of Texas at Austin (UT), and then worked at the Federal Reserve in mortgage and banking resources for four years. After that, she moved back to Austin to work as a research analyst at UT before moving to NC State to begin a PhD in statistics. After finishing her PhD, she moved to San Francisco and joined YouTube, where she’s currently Senior Data Scientist.
On working as a data scientist at YouTube
Being a data scientist can mean lots of things at many different companies. At Google, it means you’re an in-house statistician. Up until earlier this year, I was a lead on YouTube Comments, where I worked on comment quality and rankings. My job there was to help design experiments to test features, make sure we’re gathering the data we need, and interpret results and data from those experiments. Recently, I moved to YouTube’s racial justice and equity team. Now I work on more horizontal problems and get to look at different parts of YouTube, such as search results or recommendations.
What was your day-to-day like while you were working on YouTube Comments?
I worked mostly with software engineers who were working on the comments. I would have meetings with software engineers or product managers about new ideas or questions that they have about something they’re working on. For example, if they wanted to tackle some problems, we’d meet to brainstorm how to do that. I also did a lot of technical work myself. For example, I write a lot of code. I’d say it was about 50 percent technical work and 50 percent meetings with people.
What strengths and skills do you need to be a good data scientist?
Communication! A lot of people think that this is something that you’re born with but it’s really something you can practice and it’s a huge part of any kind of interdisciplinary work. On a daily basis, I have to talk to everyone from software engineers to marketing specialists. These are people who have vastly different backgrounds from each other and also from myself. I have to make myself understood by all of these people. The best data scientists that I know are all excellent communicators.
Some ways you can develop good communication skills are to teach and tutor. Of course, you also need math and coding skills… but presentation and communication skills are absolutely critical.
On her favorite part about working at YouTube
My favorite part is that I get to touch the everyday lives of millions of people every day. For example, during the pandemic, everybody learned how to do new things on YouTube. You could look up how to cut your own hair… or how to bake bread!
I also see experiments that I got to work on that get written up in the technical press… and it’s just crazy! It feels very tangible to me.
Can you share what the interview process is like at Google?
First, it begins with a meeting with a recruiter. They’ll go over your interests and see if there are openings in roles that are a good fit for you. Next, you’ll meet with a technical lead. They’ll discuss all the statistical and computing basics to make sure you’re a good fit for the role technically. Then, there will typically be onsite interviews (though currently, they are all virtual). During this time, you’ll typically have five technical and four non-technical interviews. Each of these typically lasts about 45 minutes.
Afterwards, all your information gets combined into a packet. This packet will also include previous interviews and everything gets sent to a hiring committee. Afterwards, you may meet with potential managers from several teams. Finally, you make a decision about which team you want to join. It’s a long process so it’s a good idea to start the process early.
For interviews, will you be asked to code live? How do you prepare?
I ask interviewees if they want to use R or Python. You can choose the language that is more comfortable for you. I’m not necessarily looking for the right. What I’m interested in is how you consider a problem, your problem solving skills, and your thought process.
There are generally multiple ways to approach a problem. So it’s really an opportunity to see how you approach it. Generally, you also won’t be given enough information to answer the problem. So it’s also an opportunity to see what questions you ask. It’s a great idea to think out loud to let your interviewer know what you’re thinking about.