The Big Data bible from Stanford: MapReduce, Spark, recommendation systems, PageRank, locality-sensitive hashing, Large scale machine learning and mining social networks/streams all explained clearly with real algorithms you can code today. 500 pages of pure gold.