A research director at the CNRS, he began his career by making decisive contributions to the Baum–Connes conjecture in the theory of
operator algebras. He then turned to the Langlands program (see *
Tangente*
149), a field in which his brother had distinguished himself. He went further still, using a highly general ("many-legged") version of the
shtukas introduced by Vladimir Drinfeld in the 1970s to tackle the Langlands program dynamically: in a sweeping two-hundred-page paper written in French, he found a
"wonderfully simple and direct argument" that finally explains why the correspondence conjectured by Langlands
"is not a miraculous consequence of complicated calculations" and why
"it has to be true", according to Richard Taylor, who helped prove Fermat's Last Theorem with Andrew Wiles in 1995.