Yes. it can. Scikit-learn supports algorithms that can be used to build your own recommender system. To scale your algorithms you can always use hadoop map/reduce to scale.
The best documentation I found is the Mahout in Action book (http://manning.com/owen/) while reading the source code in parallel.
Also you probably don't need to run this on a Hadoop cluster unless your data is too big to fit on one single machine.