The Axiom [1] developers created a special purpose dependently typed language to capture mathematical abstractions: SPAD/Aldor [2]. Maxima uses Lisp. [3] These packages contain many man-years of work.
Julia is JIT-compiled, a Lisp (under the hood) and (somewhat) dependently typed.
Is that enough to port (or even transpile) modules from the big open source CAS without an entire rewrite? Is there enough similarity between CAS to make "foreign" modules even a remote possibility?
Otherwise, Nemo will likely not achieve a great unification of math packages, since the required effort goes way beyond the resources of a small dedicated group.
Since Julia is about speed, how does this compare to Sympy speed?
This is a fantastic collection of software into what seems to be a well organized and well documented package.
There's a little bit more information in Fredrik J's blog post: http://fredrikj.net/blog/2015/09/finding-nemo/
If it is built on PARI how does it improve upon it?
Personally I want to know how to define algebraic number classes. I attempted and failed write my own algebraic number types in Python.
Holy shit! Those benchmarks look absolutely incredible, outperforming Magma by a factor of 5 on average. Incredible work. This is the kind of work that makes Julia really look like the next step for scientific computing.
http://nemocas.org/benchmarks.html