Researcher of the Month

“I feel lucky to work on answering questions that are not only deeply interesting and fun to explore, but also profoundly impactful — shaping how we solve problems across so many areas of computation, from advancing machine learning systems like ChatGPT, to strengthening cryptography and making algorithms faster and more efficient.”

As a theoretical computer scientist, Josh Alman is dedicated to exploring the possibilities and limitations of computation through the lens of algebraic algorithms. Building on a lifelong love of math and puzzles, Josh has emerged as a leader in advancing the efficiency of computational processes. His research focuses on using mathematical tools to create faster algorithms.

In areas such as matrix multiplication — a cornerstone of modern computational systems — Josh’s contributions are particularly impactful. One compelling example of his work’s significance is its application to large language models like ChatGPT. When users ask ChatGPT a question, the system converts the input into an algebraic object called a vector. The model then performs extensive algebraic operations on that vector, such as matrix multiplication, to produce a coherent response. In other words, “When you ask it a question, in the background it is actually doing lots and lots of algebra — and then spitting out an answer at the end,” Josh explains. His research aims to optimize these algebraic operations by making them faster, using less energy and hardware, and ultimately enhancing the performance of AI technologies.

Josh describes his work as a unique intersection of intellectual curiosity and real-world impact. “I feel like I get to work all the time on these really interesting, fun puzzles — but simultaneously, I am hopefully making progress on answering important questions that will really help people in a lot of different domains of computation,” Josh shares. He feels fortunate to work in a field where answering questions he finds fascinating can also have significant, positive impacts on the broader world of computation. A lifelong learner, Josh is also deeply curious about neighboring areas of computer science, and he enjoys developing new connections between them to apply novel insights and techniques to his work.

In addition to his professional achievements, Josh also enjoys spending time with his family, especially exploring New York City together. From visiting parks and museums, to discovering kid-friendly restaurants, he embraces the city’s vibrant opportunities for family fun. In his personal time, he indulges in his love of video games. His enthusiasm for League of Legends even inspired him to write a research paper during his PhD on the computer science underlying the game.

Josh also has a long-standing passion for puzzles. He says, “I really enjoy solving puzzles, and participating in and making something called ‘puzzle hunts,’ which are collections of puzzles where you get together with your team and race other teams to solve all the puzzles as quickly as possible.” For over a decade, with friends he has been creating “puzzle hunts,” which he releases online for others to solve.   

Josh is quick to credit much of his success to the collaborative nature of his field. “Mathematics and theoretical computer science are very collaborative areas, and I work together with so many great students, postdocs, and faculty at Columbia and around the world,” he says. “A lot of my research successes really should be attributed to them.” This sense of gratitude reflects his belief in the power of teamwork and mentorship, further cementing his role as a pioneer in theoretical computer science and an inspiration to the next generation of researchers.