Emmanuel Candès has built a career around a radical idea: you do not need all the information to find the truth.

As the Barnum Simons Chair in Mathematics and Statistics at Stanford University, Candès works where mathematics, statistics, computation, and engineering meet. Much of his life’s work circles a deceptively simple question: how much can we know when much of the evidence is missing?

His most celebrated breakthrough, compressed sensing, overturned conventional thinking about how information must be collected. High resolution had long demanded enormous numbers of measurements. Candès, working with Terence Tao, Justin Romberg, and others, showed that when a signal has underlying structure, it can be reconstructed with astonishing accuracy from a fraction of the expected data.

The mathematics quickly escaped mathematics. In MRI, compressed sensing opened a path to faster scans by reconstructing images from less acquired data. The consequences are human. Minutes matter when the person inside that machine is frightened, in pain, struggling to breathe, or a child trying desperately not to move. The same ideas have reached astronomy, communications, radar, and other fields where measurements are costly or difficult to obtain.

Candès carried this thinking into matrix completion. Given a vast dataset with most of its entries missing, could the whole still be recovered? He established the conditions under which a surprisingly small scattering of observations can reveal the larger structure with mathematical certainty.

More recently, Candès has taken on another fundamental problem: how do we know a scientific discovery is real?

Modern genomics, neuroscience, medicine, and machine learning can generate millions of variables at once. Search through enough data and chance begins to masquerade as discovery. Candès and his collaborators developed Model X knockoffs, a method that creates synthetic control variables and pits real variables against them, helping separate genuine signals from statistical mirages while controlling false discoveries.

Born in Paris in 1970, Candès studied at the École Polytechnique before earning his doctorate in statistics at Stanford under David Donoho. He later became the Ronald and Maxine Linde Professor of Applied and Computational Mathematics at Caltech before returning to Stanford, where he eventually chaired the Department of Statistics.

His honors include the Alan T. Waterman Award, a MacArthur Fellowship, election to the National Academy of Sciences, the Princess of Asturias Award, the IEEE Jack S. Kilby Signal Processing Medal, and the 2026 Shaw Prize in Mathematical Sciences.

Across his work runs a question that feels almost philosophical: how much can we know when the picture is incomplete?

Because the picture is always incomplete. Nature does not hand us clean datasets or finished answers. We get fragments. A few measurements. A faint signal in a great deal of noise.

Most of us look at what is missing and see the limits of what can be known. Candès looks at what remains and asks if it might be enough.

Again and again, he has proved that it is.