By background I'm an embedded systems guy. In college I spent years focusing on microcontroller systems and FPGA hardware. I'm not going to spend the whole post reminiscing about the "good-ole' days" when I still worked down on the metal, but I do want to express my particular fondness for this next GSoC project idea: The Parrot-on-RTEMS port.
Jonathan Leto has been spearheading the project so far, but it's become clear that there are some major deficiencies in Parrot's cross-compilation facilities to enable Parrot to build and run properly on a real-time OS like RTEMS. Also, there are other problems with Parrot's algorithms that prevent it from meeting strict real-time deadlines.
In a message to the mailing list tonight Johnathan put out an official-looking call for interested participants, and also advertised an extremely informational page on the RTEMS wiki.
I will let the RTEMS page discuss most of the good details about what is needed, so I will cut off this blog post here. But if you're an interested student and want more information about this project, head on over to the RTEMS wiki and get in touch with Jonathan Leto.