By Jonathan Shaw
JELANI NELSON LIGHTS UP when he talks about algorithms. The soft-spoken assistant professor of computer science is a rising star in a field made vital as data proliferate exponentially faster than the growth of computational power or storage. Algorithms, well-defined procedures for carrying out computational tasks, speed the way to answers. Nelson has a knack for speed: online, where he is known as “minilek”—a handle chosen in youth when he was growing up on St. Thomas, and derived from the name of an early ruler of Ethiopia, whence his mother hails—he has excelled with equal ease in coding competitions and typing contests (topping out above 200 words per minute). Though he is a theorist now, solving real problems quickly “cements the concepts in your mind,” he says. Borne of that conviction, every homework assignment in his undergraduate course Computer Science 124, “Data Structures and Algorithms,” includes an algorithmic programming problem. His own student years were spent practically next door, at MIT, where he majored in mathematics and computer science, and remained to earn a Ph.D. in the latter field. He came to Harvard in 2013 after postdoctoral research at Berkeley and Princeton’s Institute for Advanced Study. Nelson’s specialty is “sketching,” an approach to dealing with problems in which there are “too many data in the input.” He figures out how to create compressed, often exponentially smaller, versions of datasets that nevertheless retain useful, accurate information. His proofs defining the limits of such approaches have illuminated fundamental questions, some of them unanswered for decades. Though he is humble and quiet, his colleagues are less reserved: they call him “simply brilliant.”