Interview with Skyler Wu, May 2024 Concentrator Prize Winner

In 2024, alum Skyler Wu received the 2024 Department of Statistics Senior Concentrator Prize at our Commencement Celebration on May 23, 2024. Wu graduated with an AB in Statistics and Mathematics (Joint) and an SM in Applied Mathematics. He was selected for the prize by the department based on the high quality of his coursework and thesis. After completing his degrees at Harvard, he started the next chapter in his statistical journey by joining the Statistics PhD program at Stanford University.

To learn more about Skyler Wu’s challenges and triumphs in statistics as well as his thoughts on what he will miss most about the department, we had the following conversation with him, edited and excerpted below. Congratulations, Skyler!

Was there an experience that made math or statistics personally engaging?

Wu: Honestly, I entered college with a feeling of imposter syndrome (and COVID didn’t help!) because everyone seemed to be on such a different level. When I heard about students placing into Math 55 [honors abstract algebra] and representing their countries in math competitions during high school, I wondered whether I belonged.

The experience that changed my perspective and gave me more confidence was taking Stat 110 Probability with Professor Joe Blitzstein. Although I originally signed up for Stat 110 because I figured I should fill a prerequisite (Stat 110 is a prerequisite for many STEM concentrations) while being cooped up inside, to my surprise, I ended up loving the class! Interacting with Professor Blitzstein and amazing teaching fellows (TFs) like Rachel Li (AB ’23), Ginnie Ma (AB ’23), and Yash Nair (AB ’22) made me feel like I belonged and had a seat at the table.

During freshman spring, my interest in statistics and math really cemented when I took CS 181 Machine Learning. There was a lot of buzz in the air about machine learning, so I was curious and wanted to explore it. I quickly realized that what people call “machine learning” or “artificial intelligence” is really applied statistics and applied math. This core realization compelled me to pursue a path of studying statistics.

What initial memories do you have of getting to know the Stats Department?

Wu: As I mentioned, one of the most formative experiences was Stat 110, even though it was over Zoom. During COVID, we weren’t allowed to interact in person much, but through Stat 110, I was able to meet a lot of phenomenal TFs virtually—there were basically office hours around the clock—who made me feel supported and welcomed in the Statistics Department. Of course, Professor Blitzstein, as the instructor of Stat 110, really reinforced this positive experience. I even told Professor Blitzstein that, before coming to Harvard, I held an image of faculty far removed, sitting in lofty ivory chairs. But Professor Blitzstein didn’t match my image at all; instead, he directly interfaced with undergraduates in a manner that was very approachable, friendly, and supportive. He helped me transition into college and the stats department.

What motivated you to pursue a Statistics Concentration?

Wu: At first, I thought I would pursue a career in science policy and diplomacy. I wanted to combine technical skills with a background in policy to address big global challenges. Eventually, I realized that nearly every discipline requires the tools to work with data, understand uncertainty, and predict future patterns; they require the common backbone of statistics. I first heard the saying “The best thing about being a statistician is that you get to play in everyone's backyard" [attributed to the statistician John Tukey] from some of my professors, and it really hit home for me. Regardless of your field, statistics is a common denominator, which made it a darn good option to pursue as a concentration!

I’m also very interested in machine learning, but I like approaching it from a statistical angle. When examining a machine learning model from a statistics perspective, I appreciate that you need to be explicit about the assumptions that you make about the model and its data. Another reason why the concentration was a good fit was because I feel naturally drawn to questions such as, “To what extent do we believe this prediction?” and “What is the level of uncertainty about our results?”

In addition, I like the beauty and unity that statistics offers. For example, in Stat 230 [Multivariate Statistical Analysis], Professor Sam Kou (also my thesis advisor) introduced some modern statistical techniques during the last third of the semester, including principal component analysis, canonical correlation, and critical angles. Professor Kou emphasized that all these statistical tools derived from the same exact underlying theorem, which reveals the kind of beautiful unity and organization in statistical thinking that I so admire.

Did you encounter challenges (personal or academic) during your studies? How did you overcome these challenges?

Wu: One of my challenges was taking CS 181 Machine Learning during my freshman spring. Initially, I felt like I bit off more than I could chew; I barely knew python and thought maybe I should quit. However, Yash Nair, who was a teaching assistant for the course, encouraged me to stick with it and helped me rethink how I approached studying and problem-solving—he really helped me to become more resilient. I’m so glad I listened to his advice because I ended up enjoying the course. I even ended up TA’ing for the course the following year, and by junior year, I was one of two co-head TFs!

Another challenge came during sophomore fall, when I took Stat 210 [Probability I], my first graduate-level course, with Professor Blitzstein. Professor Blitzstein must have known that I needed some encouragement because he shared an article about the “Ben Franklin method,” which uses the example of Ben Franklin improving his writing as a kid to argue for the importance of deliberate practice. The article reinforced the idea that learning doesn’t just take place if you lock yourself in a room; rather, it’s about identifying the parts you don’t get and then structuring practice in that area. Long story short, this message stuck with me and the strategy worked out; I gained confidence and finished the semester strong.

How did you start collaborating with Prof. Sam Kou and how did you select your thesis topic?

Wu: Professor Kou, my thesis advisor, has been one of the most influential mentors in my time here. I connected with Professor Kou through Professor Mauricio Santillana at the Harvard School of Public Health. While working on a disease forecasting project, I expressed an interest in pivoting towards more methodological work, and Professor Santillana suggested Professor Kou.

When I took an applied math course [APMTH 216 Inverse Problems in Science and Engineering] with Professor Michael Brenner, I started to ask: “Could we use Bayesian approaches to analyze systems as complex and unpredictable as the Lorenz system?” My proposal to Professor Kou was to look at how his lab group’s Bayesian method (called MAGI—Manifold-constrained Gaussian Process Inference) would work under the Lorenz system, which is used to model complex phenomena like climate.

To give a brief introduction to the Lorenz system, it is a dynamical system that has a bunch of variables that evolve over time and interact with each other. Some examples of variables within a Lorenz system are: 1) a disease-infected population vs. a recovered population and 2) a predator population vs. prey population. For these types of variables, it’s very difficult to predict how they will evolve in the future; even if you are off a tiny bit numerically, it can cause your prediction to go awry.

For my senior thesis, we built upon the MAGI method by creating a version called Pilot Magi, which addressed some of the numerical instabilities in MAGI. By building Pilot Magi, we met our goal of building a tool that anyone working with dynamical systems could use. For example, in a disease forecasting setting, researchers might want to ask for a prediction of the rate of infection over time. Also, our method could help by taking noisy and sparse data and reconstructing what the ground truth would have looked like without this noise.

What do you value the most about your experience in the department and at Harvard? If you had to select a word to encapsulate your stats experience, what would it be?

Wu: “Inspired” is the word that best captures my experience at Harvard. There are two big takeaways for me about Harvard and the Stats Department; my experience has provided me with a sense of purpose and meaningful connections to people. From interacting with various role models at Harvard and in the department—professors, grad students, and upper-class undergraduates—I have graduated with a real sense of purpose. I would like to become a professor of statistics and machine learning to advance cutting-edge research and to democratize these tools and knowledge so that they are available to the public as much as possible.

Equally important to me are the people that I met every day at Harvard who inspired me. Many of my classmates were graduate students whom I considered older academic siblings, and I looked up to faculty and my undergraduate friends. These classmates, mentors, and friends were wildly different in their passions and interests, but they all wanted to use their talents to make a positive difference in the world. The good part about being in this digital age is that, while I have left Harvard, these lifelong mentors, friends, and role models will stay with me.

What are you excited to pursue this fall? Describe some of your career, academic, and personal aspirations and plans.

Wu: Over the summer, I visited family in China, which was wonderful. On a fun note, I finally had access to a kitchen at home and enjoyed cooking for my family.

In the fall, I’m starting my PhD in Statistics at Stanford University. A short-term goal is to finish up research work with Professor Kou, with the aim of completing a journal submission. My long-term goal is to become a professor, like many of the incredible mentors I’ve had at Harvard, especially Professors Blitzstein and Kou. I’m excited to continue growing as a researcher and teacher and further pursue my aspirations of making statistics and machine learning accessible and impactful.