I used DipTrace to lay it out which is restrictive in pin-count instead of sq-in.
If you can wade through the schematic for any of the standard Arduino boards there's a lot of copy/paste here.
I am working on a robotics platform that requires a lot of multitasking so the multi-CPU design made a lot of sense. The other cool thing is these boards can be daisy-chained or connected to other I2C devices.