Right now we are doing a lot of work to support Spark / Hadoop and several associated libraries / frameworks - SystemML, MLLib, Mahout, etc. Eventually we'll add support for TensorFlow, CaffeOnSpark, Singa, etc.
Most of what we've done so far is around provisioning (we're building on AWS for now), but we've done some work on the service APIs for training models, making predictions, etc. Adding PMML support in in the cards for "real soon now" as well. Beyond that we have a whole laundry list of features to add.
Not quite ready to launch yet, but things are taking shape. I'm really hopeful that we can at least hit our closed beta launch sometime in the next month or two.
Oh and the other thing is, our platform itself will be Open Source software, and will be available for deployment into private environments if you don't want to use our hosted service.