check out cs224d and cs231n (both stanford) , there's another course by university of waterloo https://uwaterloo.ca/data-science/deep-learning , you will find lecture videos for all on youtube
Slightly off topic but for anyone who is taking this course ...are the materials only related to NLP or are the techniques much more broadly applicable to other areas of deep learning (cursory look of the syllabus suggests this but would be great if someone who is actually taking this course can comment)
I am working my way through Alex Aiken's Coursera compiler course, almost 3/4th done and really liking it till now.
There was another resource posted on HN earlier which takes you through building a compiler for lisp using C.
http://www.buildyourownlisp.com/contents
https://pdos.csail.mit.edu/6.828/2014/schedule.html
The mit website pretty much has all the information open, which you can use to go through this course at your own pace. The lectures(most if not all) are on youtube, search for 6.824
Maybe you are correct, Hbase is not easy to configure for sure but in terms of scalability (sheer number of row keys you can dump) there are not a lot of other components which can compete.
Although theoretically this abstraction can live on top of any backend which has facilities for fast prefix scans.
The primary motive for writing this was the fact that Hbase gives a fairly cheap(hdfs) backend storage and near linear scalability. Also most of the schema is encoded in row keys in such a way that hbase prefix scans can be leveraged to serve traversals across billions of nodes/edges.