The programming language is fixed. Target GPU is Kepler and/or Maxwell. The brute-force Shannon algorithm (Type A) should be implemented (in a nicely commented code). The ELO strength is not important, but efficiency considering the given algorithm is yes. Including all the chess-rules is essential. Graphical interface is not needed. UCI command line interface is essential.
The selected person should work collectively with me on the details. I have programming knowledge, but not of CUDA. Testing environment: both Linux and Windows.