Congress shut down, so she became a data scientist at NetflixLearn how Becky Tucker overcame a setback in her academic career, what she does at Netflix and the importance of listening with empathyAndrei LyskovBlockedUnblockFollowFollowingMay 1Netflix has revolutionized the way we watch and has turned subscription-based streaming content into the norm.
With over 139 million global subscribers and a US market share of 51%, Netflix is a dominant player in this space.
Becky Tucker has been with Netflix for almost five years.
She’s currently a senior data scientist on the Studio Science and Analytics team.
In this article, we talk about her transition into data science from physics, her work at Netflix and the importance of listening with empathy and humility.
Pivoting Into Data SciencePrior to her career in data science, Becky was fascinated with physics and the wider cosmos.
She did her undergrad in physics at Georgia Tech, with a focus on astronomy.
Spurred on by her enjoyment of teaching and learning, she pursued a physics PhD at Caltech with the Observational Cosmology group.
She joined the group as an experimentalist, doing lab work and data analysis in both Matlab and Python.
However, by her fourth year, she started to grow disillusioned by her job prospects in academia.
The type of work that interested her meant there would only be two or three tenure-track academic positions a year, and they typically had very stiff competition.
To make matters worse, in her final year of grad school, a government shutdown caused the Office of Polar Programs to cancel an entire season of instrument launches, which created a two-year delay in the launch of her experiment.
As a result, she could either stay for two more years to see her project to completion or pivot her career into something new.
Since her entire lab’s equipment was stuck in New Zealand, and it would be several months before they could get it all back, she had a lot of time on her hands.
This empty space allowed her to revisit her doubts about academia, so she started to explore some typical alternative careers for a physics PhD, including quantitative finance, aerospace, defense, scientific consulting and the relatively new field of data science.
From her reading and discussions with individuals in the field, she realized that data science was the perfect fit for her.
While she had some experience programming and had worked with neural networks in a few of her internships, the field was quite foreign to her.
Luckily she came across the Insight Data Science program, a program geared towards PhDs in STEM fields who want to transition into data science.
Given the empty space in her schedule, she was able to get her advisor’s blessing to participate.
Going into the program she had good Python skills for a physicist, but poor skills for a programmer.
She also had no knowledge of databases or SQL and limited knowledge of machine learning and a number of other concepts needed to be an effective data scientist.
However thanks to her training in physics, she was less intimidated by hard problems and had already cultivated the skills of teaching herself difficult concepts.
This mindset was captured for her in the German phrase “Sitzfleisch”, which roughly translates to sitting flesh, or the ability to sit in a chair long enough to work through a difficult problem.
Towards the end of the program, the class had an opportunity to do a roadshow, where they show off their personal projects at different companies within the constraints of a two-minute presentation.
Becky’s project was OnLocationMap, a web app that identified and plotted locations where movies were filmed using Python and MySQL.
She was already a huge film buff and, as a result, the project was one which she felt genuinely interested in bringing to life.
Between her personal interests in film, and budding data science skills, a role at Netflix seemed like the dream role.
After pitching her project at their offices, she was able to secure an interview, which led to an offer to start in three months.
This gave her enough time to go back to graduate school and finish her thesis (using calibration and testing data, rather than flight data), allowing her to graduate with her PhD and start her new role.
Reflecting on her time at Insight, Becky was thankful for the experience it provided her.
The most valuable thing she got out of it was learning about the things she didn’t know that she didn’t know.
Specifically, what industry looked like, how people think about business metrics, as well as the jargon and skills required to be successful in a data science role.
It also affirmed her decision to go into tech, and leave behind the world of academia.
Data Science at NetflixFresh out of graduate school, Becky joined Netflix in their Los Angeles office on the Data Science and Engineering team.
At the time the majority of the company worked in Los Gatos, so the Los Angeles office was quite small.
Her team consisted of 12 people, with roughly 250 people in the LA office in total.
She found the transition easy, thanks in part to the amazing and supportive people on her team that helped her ramp up.
At no point did she ever feel bad about not knowing something, and the team actively helped her find projects in the first few months.
During the early days, a lot of her work surrounded the analysis of Netflix Original TV series, which at the time consisted of three shows.
The work on these projects was a sharp contrast to her work in academia, where it was difficult to see the immediate impact of her work.
She soon found herself in a senior role, where she currently splits her time between analytics and predictive modeling.
The type of work she does often varies, and at any given time she may be working on three medium and one small projects.
Of these projects, 50% may pay off within the quarter, while the other half may have deliverables that extend beyond the quarter.
While the type of work she does is often collaborative with shared codebases, most of her projects are self-contained.
One of her favorite aspects of the Netflix culture is the focus on context, not control.
While her manager and stakeholders can give her context of what’s going on in the company, and what they think may be important, it’s up to her to decide what is the most important thing to be working on.
This typically means prioritizing projects that have the potential for the most business impact.
Advice for OthersWhen asked what advice she would give to herself five years ago, she provided these pieces of advice:Don’t neglect the importance of software engineering in data science.
If you think that the code you’re producing will live beyond a few quarters, invest in building robust code.
This means adding comments, making code readable, and understandable for future maintenance.
While machine learning can be an important aspect of her job, her role as a data scientist is as much a business role as a technical one.
As a result, it’s pivotal to your success to be able to listen with empathy and humility to your stakeholders.
An important reframe is that your work should be helping people who are already experts to do their job better, not telling them how to do their job.
Keep learning!.Everyone is different, but she has found that one of the best ways to learn is to buy textbooks and work through them by dedicating 45 minutes each morning.
Being able to go through the fundamentals of a textbook and apply it to work is a great way to solidify the concepts.
(She recommends Elements of Statistical Learning and Doing Bayesian Data Analysis)When communicating with stakeholders, learn how to use metaphors and analogies to make your points easy to understand.
This strategy has paid off immensely when trying to explain the rationale for why something may or may not work to non-technical stakeholders.
Data Minds is a series that profiles professionals working with data.
In this series, you’ll learn about their story, day-to-day, and advice for others.
Previous interviews include data scientists from Red Bull, Open Door, and Snapchat.
.. More details