Well, you're not employed by DARPA, you're a professor at your university and DARPA gives you a grant. By that account probably a few thousand or even ten thousand "researchers" (processors, post docs, grad students, undergrads) working on DARPA projects. (I make that point because a lainon up thread said "working at DARPA" which is a common misconception.)
These projects are pretty mundane, though. I think the DARPA grants I've been near have been about program verification and security (but boring stuff). DARPA funds about half of American computer science. They're just another NSF.
If you're a career mathematician, federal payroll might be very competitive, but the pay grades you live in as an NSA developer are crap compared to even other contractors in the DC area. If you work for the NSA you're basically dooming yourself to a life of drudge work you can't talk about, long commutes on the BW parkway or worse, I-95, and a medium sized house in Colombia. Most developers say fuck that when they could live in DC or NYC, get paid more, and have better quality of life. The people who work for the NSA either want federal pensions/a government job (overtime, no real management, no real budgetary pressure), or are fascists.