First of all to contribute or to work on some projects on Machine Learning, one needs to have stronghold on concepts for which going through theories is necessary.
As mentioned by @byteace, access to research projects or ideas is difficult, but from my side I would like to suggest that if you do find difficulties in understanding papers of top conferences, or if you face trouble in sorting them out ,then do go through the coursework of various universities, which do have assignments, some links also to projects submitted by students (which you may also find out by visiting their github profiles ). They do have links to github classrooms, piazza forums and many more. You may be able to get access to a lot of resources of your interests if you explore through them.
You may visit sites like :- http://cs181.fas.harvard.edu/, https://www.seas.harvard.edu/courses/cs281/,
cs231n.stanford.edu,
https://www.cs.ox.ac.uk/people/nando.defreitas/machinelearning/
and many more…
You just need to explore them, visit through the assignments provided, understand the codeflow and try them out yourself.
You can also go through some of the problems hosted on kaggle and go through the corresponding kernels provided to understand how they are solved.
But before proceeding to all these, please do gather as much theoretical knowledege as you can, to have clear view in this field ahead.
