Grad Spotlight: Sophia Pi

Sophia Pi explored several research directions before ultimately rooting her curiosity in the foundations of AI and machine learning. She is motivated by research questions including: when do individual learning systems—or ensembles of many agents—achieve "nice" outcomes (e.g. training efficiency, trustworthiness, robustness, and collaboration)? What theoretical guarantees can we derive about the behavior of these systems? And how can we use mathematical tools to develop a principled science that will let us scale AI/ML systems in ways that align with human interests, improve social welfare, and utilize resources efficiently?

Graduating this month with a bachelor’s degree in computer science and a joint major in the mathematical methods in the social sciences (MMSS), Pi was advised her first year by Professor Miklos Racz, her second year by Professor Han Liu and the MAGICS Lab team, and her final two years by Professor Ben Golub.

She also collaborated on independent projects with her peers, including building an electronic medical record system startup called Alpime Health through The Garage at Northwestern’s Residency Program with Northwestern Engineering’s Tahira Grewal and Isaac Meite. The team won first place in the VentureCat 2024 Life Sciences and Medical Innovation Track competition.

When she wasn’t in the lab or classroom, Pi was often at the rink. For three years, she has been a member of the Northwestern University Synchronized Skating Team, The Purple Line. Part of the University’s Figure Skating Club, the team competes at the Open Collegiate level.

We asked Pi about her experience at Northwestern Engineering, impactful interdisciplinary experiences, and her advice for current students.

Why did you decide to pursue the CS major at McCormick?

I went into college knowing essentially nothing about what I wanted to do. I knew I liked math—not so much for the pure rigor of it but because I loved the way that people used mathematics as a language. During my first year at McCormick, I took a collection of random classes across engineering, CS, and economics and I came to understand math as a medium through which we could tell precise and beautiful stories about the systems, machines, and abstract ideas that make up our worlds—be it mechanical, informational, or sociological.

CS was an exciting nexus where the mathematical aesthetics I'd grown to love powered the most cutting-edge, timely intersections between computation and fields that spanned everything from cognitive science to law. Ultimately, the infectious enthusiasm that students and faculty members (in particular, my first-year summer research adviser, Professor Miklos Racz) had for their work solidified CS as my choice of major. They were building ideas that would lead to technological innovation and governance in ways that I couldn't stop thinking about, and I wanted to be a part of that effort.

How did the McCormick curriculum help build a balanced, whole-brain ecosystem around your studies in CS? Any particular course highlights you'd like to share?

I've come to appreciate how deeply the ethos of interdisciplinary learning and work is embedded into the curriculum. Design Thinking and Communication (DTC), the two-quarter first-year design sequence that all McCormick students take—and a course that I, as a more theoretically-inclined student, somewhat begrudgingly added to my schedule—was one of the courses that really set the foundation for my four years here. Over the course of a combined twenty or so weeks, I learned just how vital it is to incorporate insights from different domains of engineering (and beyond). The knowledge didn’t come from pre-written lectures or assignments, but through finding unexpected ideas during brainstorming sessions with students from other engineering majors, through interacting with clients from diverse backgrounds, and through constantly being challenged to understand problems from different user scenarios.

Numerous other courses and programs played an integral role in realizing this "whole-brain ecosystem,” including COMP_SCI 397, 497: Innovation Lab: Building Technologies for the Law led by Professors Kris Hammond and Daniel Linna, where my team and I developed an online game to simulate moot court for law students.

What does 'human-centered design' mean to you as a developer, and how has it impacted your approach to coding or building systems?

The systems we build should be motivated by people, not from technology. Particularly in this new era of AI, it can be very tempting to jump on all the magical abilities that we could potentially take advantage of and simply find people who would want our product. Analogously, in my theoretical research, it can be tempting to use the coolest new mathematical tool I learn about and try to retrofit a problem that induces that tool. Just as my experience in the McCormick curriculum has emphasized the importance of starting with an empathetic, human-centered approach to system development, I've realized that the research work I am most proud of almost always comes from building first principles based on an organic problem.

What skills or knowledge did you learn in the undergraduate program that you think will stay with you for a lifetime?

The skill of being able to pivot quickly between different domains and constantly having a subprocess churning in the back of my mind looking for connections across seemingly disparate fields will always be something I value. On the interpersonal side, one of the most valuable things my time at Northwestern has given me is a certain degree of shameless optimism that I sincerely hope will remain with me for life. It took quite some time for me to finally put aside my fear of social awkwardness in networking, cold emailing, and talking to people at seminars, but once I did, it became one of the most fruitful qualities I've cultivated. I got access to opportunities I would have been completely oblivious to otherwise, I learned about connections that fundamentally changed the way I thought about my work, and more importantly, I found people who have become some of my most cherished friends and mentors.

What's next? What are your short- and long-term plans/goals in terms of graduate studies and/or career path?

I'm excited to be starting my PhD in computer and information sciences at the University of Pennsylvania this fall! I'll (tentatively) be working on trustworthy AI and multi agent systems and am very lucky to be advised by Professors Aaron Roth, Michael Kearns, Surbhi Goel, and Meena Jagadeesan. Long term, I'm keeping my options open— this is simultaneously a very exciting and very unprecedented time to be working in this field, so I am excited to continue exploring opportunities as they come up.

What advice would you give to current or incoming CS students?

Talk to people! Northwestern truly is a uniquely welcoming place with an incredibly collaborative community, and your peers, mentors, and professors (arguably more than the content of the courses) will be your biggest source of opportunity, inspiration, and support. Nearly every single opportunity that I had during my four years here was acquired by making connections with people: my peers in my calculus class, faculty I cold-emailed, speakers I chatted with after their talks, or just random people I bumped into through the unpredictable Brownian noise of life. Four years can feel like an impossibly long time, but finding the right people will make you wish it were longer.