Courtney Paquette, Ph.D., an assistant professor in McGill University’s Math and Stats Department, traces a path into research that starts far from the usual origin story. Speaking with Saura Naderi, UC San Diego, Paquette describes earning an undergraduate degree in finance before being pulled toward the applied math questions hiding inside real systems, including operations research and the optimization problems behind logistics. From there, she keeps following the thread, taking more mathematically focused courses in areas like probability and numerical analysis, until she lands in a pure math graduate program feeling “completely unprepared.”

That early shock becomes part of how she defines research itself. Paquette connects the grind of getting through demanding coursework to the day-to-day reality of original work: long stretches where things do not work, where you have to stay creative, and where progress means learning to “work in this unknown.” She returns to the idea that growth requires being comfortable with being uncomfortable, and that the struggle can build the skills that make a stronger researcher.

She also highlights how much mentorship and access matter. A professor notices her in a numerical analysis course, introduces her to math research, and “pulls strings” to help her enter a Ph.D. program even without the prerequisites. Later, she describes a Ph.D. in pure math focused on optimization, and an unexpected detour into industry after receiving a job offer from Google Brain. She works there for a year and is struck by how much theoretical thinking is valued, and how much she learns from collaborating with people on the engineering side.

When Paquette turns to machine learning, her message stays grounded in practical constraints. She argues that today’s systems are increasingly limited by computing resources and energy, and that the core challenge often becomes an optimization problem: given a fixed compute budget, how do you design a neural network to get the best answer without wasting energy and resources? In her telling, that tension, between ambitious ideas and finite capacity, is exactly where careful thinking and well-designed algorithms matter most.

Watch Big Data, Better Answers: Optimization at Scale with Courtney Paquette.

For more programs like this, visit UCTV’s Data Science Channel.