In the past, I have been developing several auto benchmark and data set generators to build training sets for machine-learning based compilers. Now, I am interested to create a universal cM module (possibly combined with Alchemist) to generate multiple micro-kernels, codelets, benchmarks and data sets for systematic program and architecture analysis and optimization.

As a dummy, I started adding cm dataset create for polyhedral benchmarks ... It can be an internship or part of a collaborative project.