I am about 1/3rd through reading the thesis. I believe it's very interesting, but not 100% applicable to Scala. In particular, recursive types are used pervasively in the thesis, and the principal types property depends on them. But they are not present in Scala or in DOT. DOT has a different sort of recursive type, which is restricted to records, and the recursion goes through the self reference
this instead of the type itself.
There's some doubt whether MLsub's recursive types let too many programs pass the typechecker which would better be rejected. See Andreas Rossberg's comment in http://lambda-the-ultimate.org/node/5393.
Leaving out recursive types, the other elements are indeed very close to DOT and Scala. Maybe we can learn something from the thesis for making our constraint solving and type inference better.