Yes,
http://www.lfd.uci.edu/~gohlke/pythonlibs/
is the go-to place for Windows compiled binary Python packages (when
you can't get them via conda). They are generally packaged with a nice
installer that will ask you which Python instance on the system you
want them placed in. These days the only thing I commonly need from
there is psycopg2 (Postgres driver) which for some reason is not
available otherwise and is a pain-in-the-neck to build and install
otherwise.
Fortunately conda completely solves the Python packaging problem for
the (very many) packaged they provide (and others can be found on the
binstar community conda package sharing site). List of official
conda-available packages:
http://docs.continuum.io/anaconda/pkg-docs.html
You can also use pip to install stuff on top of Anaconda if there's no
other solution.
The combination of Python's universality (you can pretty much find
Python packages to do, or talk to, anything), the power of the SciPy
stack (NumPy, Pandas, etc.), the iPython Notebook to drive all this
power and use it for collaboration, and the Anaconda Python
distribution to make access to all of this totally painless, together
make up a system for technical computation which is unmatched at this
time in my opinion. Everyone should really give it a try. Here's an
example iPython Notebook using the public notebook viewer site that
demos the Bokeh plotting library, another another open-source package
from the same people who put Anaconda together:
http://nbviewer.ipython.org/github/bokeh/bokeh-notebooks/blob/master/quickstart/quickstart.ipynb
it shows you the sort of things you can do using the notebook system.
G.