In the late summer of 2006, I started a project to create a software package for analysis and visualization of astrophysical simulation data.
  The code ultimately came to be called yt, and it currently has somewhere in the neighborhood of 20 contributors distributed across the globe, with over a hundred people on the users' mailing list.  It's free software and open source, and it even has a set of guiding principles.  We wrote a method paper in the fall of 2010 about yt.

yt has grown to handle a wide variety of analysis tasks: halo finding, multi-variate profiling, even some light stellar population synthesis.  Its visualization capabilities have grown, too, and it now comes with a homegrown software volume renderer with a variety of mechanisms to process and output the data.  The backend has mostly been parallelized, and for some tasks it even scales to O(1000) processors.  The development team has grown, and it has an active mailing list and user base.

I use yt for all of my research -- my entire thesis was written using yt to analyze the data produced by Enzo.  Thanks to the other developers, it's grown to be a lot more than I thought it would be, and we've even extended it to work with other astrophysical simulation codes, like Nyx, FLASH, Ramses, Orion, Chombo, and Tiger.

For more info, including more about its capabilities and how to use it, see the website at .  There you can find a link to the video gallery, the blog, as well as our source code, links to our community, and so on.  We're holding a workshop at the FLASH center in January of 2012; for more info about that, we have a page with registration, and soon videos and curricula..

What I'm most excited about by yt, though, is that it's grown to be a physically-motivated way of looking at data.  Instead of loading up a simulation and throwing away lots of cartesian data, yt will load-on-demand data that has been selected to meet certain criteria: geometric, chemical, kinetic, and so on.  These data containers then get passed around and can be used as sources for chains of analysis.  The entire point is to make it easy to ask complicated questions, and I think it has mostly grown to do so in a capable and intuitive way.

There are a couple places to go to find out more about yt:

Project Statistics by Ohloh