That was George Dantzig with some statistical problems. You don't just "accidentally" prove Fermat's last theorem, at least not the way it got solved. The people involved spent their entire lives on it.

Not at all. He worked on it for years, and in secret. He went so far as to regularly publish pieces of a manuscript that he'd intended as a book on a completely different subfield, just to throw off his colleagues from suspecting.

Grigory Perelman is the one you're thinking of.