Actually, if you look at it closer and try it you will see that Wakari includes IPython Notebook - in the right pane you can chose one of three different python interpreters: python (pure), ipython or ipython notebook.

The ipython notebook is running remotely - in the cloud, not on your local machin. You have access to your explorative analytical workspace from anywhere and it's state is persistent.

This is our MVP, so there are a lot of shared features with some other existing projects. One of our "deep tech" features I don't think anyone else has is the ability to quickly and dynamically switch between different Python environments, meaning that you can easily try out new versions of libraries or test your code with new versions of Python.

We are also iterating on the overall UI and user workflow, to really facilitate the data exploration & analysis process with Python.

lmonade (http://www.lmona.de) is another (free) scientific software distribution that tries to address these problems. The technology underlying lmonade, basically Gentoo linux, can handle this.

    $ eselect python list
    Available Python interpreters:
      [1]   python2.6
      [2]   python2.7 *
      [3]   python3.1
      [4]   python3.2

Thanks for the pointer! However, note that we offer all of the Scientific Python stack built against each of these interpreter versions (and against different Numpy/Scipy versions) as well. :-)

I enjoyed using Sage for my linear algebra/probability class. Has all the light weight math functions built in, but you still have to write enough code to actually understand the problem you're solving.

Also keep an eye on https://salv.us/, I think it is/will be more or less Sage run on a large cluster.