There are many ways to contribute. You could help in making the core program better (faster, nicer, smaller...), help writing better documentation, or adding strokes for more kanji to the kanji database...
If you are interested in the actual, unstable development code of PAdict or the database generator code, both are in the CVS on SourceForge. The database generator is mostly quick&dirty C++, but seems to compile with Cygnus gcc under windows and GNU/gcc under Linux. I have only tested it on x86 machines, but the different data alignment on other systems may create corrupt output.
If you want to convert your own edict formatted file for PAdict, you can use the padict dbcreator. It can be found in CVS, but we also have a windows binary:
Don't expect too much, though, it's just a command line tool (yet)!
I have made up some database descriptions. Those are only interesting for developers. I'm not quite sure that's it's possible to understand them, but give it a try ;)
developers_notes.txt - some global database formats and descriptions
structures_tree.png - how the index trees for kana and kanji are build and stored