Understanding a number of Spark Machine Learning algorithms, e.g., Singular Value Decomposition (SVD),  their resource requriements for each step of the algorithms. Finally, propose any method or approach which either considers  algorithm properties or cluster properties to reduce the resources. Experience with Spark and Scala is important, and some computer architecture related experience/expertise is appreciated.

Spark Job Scheduling http://spark.apache.org/docs/latest/job-scheduling.html