If that's something that's still interesting to you, check out http://microcorruption.com.
I was very excited to see that for just that reason. :) I think I peaked at #24 on the day it was announced, but I'm sure I've slipped quite a bit since then.
Fundamentals are important; agreed, but the difference between knowing the theory of buffer overflows and having actually exploited one is a huge to me. Now I can say that I have actually done that, thanks to you having packaged everything up so nicely on microcorruption.com, but even that isn't 100% of the experience I would like to have, because the environment is a lot different than what I program on day-to-day.
For example, I know that I shouldn't map any memory wx (and I explained this rule in my article about JITs: http://blog.reverberate.org/2012/12/hello-jit-world-joy-of-s...), but I don't really know why. I don't actually know what an attack would look like where having wx memory around makes an exploit easier.
In either case, having that kind of memory around can only make things easier for an attacker, while by itself not being exactly a vulnerability.
And while the faculty was very strong, I sometimes felt jealous of friends from the University of Washington (the nearby Huge Research Institution) and what seemed like a more extensive CS program with more resources. I heard about some of the classes my friends took that our program didn't really have any equivalent for. I would read about some of the problem sets that sounded amazing, like a security class where you're given a VM image and the problem set is to compromise some of the programs inside the VM. Advanced curriculum like that seems difficult to develop without a lot of grad students around. A part of me has always pined for that experience I didn't get to have.
I guess it's hard to have it all.