A few years after graduating from the University of Michigan, Wendy Ku enrolled in Georgia Tech’s Master of Science in Analytics (MSA). She had been working as a business analyst and applied to the MSA program for the reason many students do: Ku was intrigued by the growing field of data analytics and its breadth of applications, and she wanted to learn more.

She’s now a senior data scientist at Getty Images (which owns iStock and Unsplash), a global visual content creator and marketplace. Her team uses natural language processing (NLP) and computer vision (CV) to train the search models Getty Images customers use to find images.

In this half of a two-part interview, Ku discusses her interest in data analytics, her work at Getty Images, and how Tech’s MSA program prepared her for her current role. In the second part of the conversation, Ku shares her experience speaking at the Women in Data Science Conference and defines “fairness in AI,” explains why diversity matters in analytics, and what she enjoys about her work.

This interview has been edited for clarity and length.

Give us a broad overview of your early academic studies and career before you entered the MSA program at Georgia Tech.

I went to the University of Michigan in 2012 for my undergraduate studies. I initially thought I would double-major in business and the arts, but when I found the arts electives to be more interesting than the required classes, I ended up majoring in business and minoring in art. 

In 2016, near the end of my studies, I took a data analytics class traditionally offered to MBA students—they held a few spots for undergrads. This was right when big data was rising to prominence, and the class introduced me to Python and SQL [Structured Query Language, used to interact with and manage databases]. It was very introductory: Our exam entailed handwriting SQL queries on paper! But because of that class, I was able to get an internal consulting job as a business analyst for a restaurant group. Most of my work there was with SQL. 

When and why did you decide to pursue a master’s degree in data analytics?

From the restaurant group I went to a cybersecurity firm, where I worked on customer-support operations analysis. I wanted to analyze text data, but SQL was just not sufficient. I was using what I knew of Python but felt like there was a gap in my knowledge. I wanted to rectify that, so I enrolled in the MSA program in 2019.

Why did you choose Georgia Tech’s MSA program?

Georgia Tech’s MSA has a long history compared to some other schools’ programs, and I liked that it’s very skills-focused and offers a good combination of different types of classes. Because my undergrad degree was in business, statistics and linear algebra were ancient history for me. In the MSA program I revisited these subjects in-depth.

I was also trying to figure out what I liked and wanted to do in data science. So the breadth of classes offered—combined with Georgia Tech’s strength in computer science and industrial engineering—drew me there.

How did the MSA program prepare you for your current role with Getty Images?

When I arrived at Georgia Tech, I didn’t know how machine learning and AI fit together with data science. During my first semester, a friend told me she had enrolled in a computer vision (CV) class offered by the College of Computing, and when I took a look at the course, it seemed fun and I enrolled in it too. 

Computer vision is a field in artificial intelligence that involves processing and understanding visual inputs, like images and videos, and I ended up loving it. It was one of my favorite MSA classes, and I decided that as part of my job search, I would prioritize any positions related to CV. That was challenging, because usually organizations hire PhDs for those roles.

Given that, how did you get your job at Getty Images after graduating from the MSA program?

One of the benefits of Tech’s MSA program is that it provides students with a travel stipend. I was able to take that money and attend the Women in Data Science conference, which is where I met my current manager. She got me in the door at Getty Images with a statistics-focused role related to A/B testing, knowing that I was interested in computer vision, and from there I eventually got more involved in CV projects.

At the beginning of our conversation, you said you had originally planned to double-major in business and art. Does your work at Getty Images let you combine your interest in art with your STEM training?

To a large extent, school was about the math and science parts of data science, but industry applications are where the arts can be brought in. Getty Images’ customers are often graphic designers or marketing professionals, and my artistic background is beneficial to my work. 

For example, when an engineer is working on an image-classification model for color themes, they might simplistically label the colors red, orange, and yellow as “warm” and blue and green as “cold.” However, a graphic designer might be looking for more nuanced differences in image styles. If a user is looking for an image of the sea and is filtering the results for warmer tone results, they probably still expect images featuring a blue sea, but a warmer shade of blue. I’m able to empathize with our users and the nuances of their use cases, and clearly understanding the use cases ultimately affects how I train the model to fit that. 

It’s easy to think of data science as objective and hard, but you’re saying there are people, with all their subjective experiences, behind these models. Does this illustrate why diversity is so important, in terms of the people who train models?

Yes—my colleagues at Getty Images all come from different backgrounds: astrophysics, psychology, bioengineering. Very few of us have undergraduate data science backgrounds. It’s our job to make sure these models are fair, and to do that, you need to incorporate the different aspects of your life and experience when working on data models. Ultimately, data science is designed by people for people. 

Read part two of this conversation with Wendy Ku, which picks up with how her work at Getty Images lets her combine her interest in art with data science.