For Professor Adam Wierman's class on ideas behind the networked world, one component was the Rankmaniac project. My team had to use Amazon’s Elastic MapReduce to quickly compute PageRanks of a large number of nodes. This was to be done in a span of 10 days with limited testing on Amazon, but unlimited local testing. The goal was to get an algorithm that would correctly rank the top 10 nodes of a graph in a short amount of time, beating the undergraduate and graduate teaching assistant teams in the CS/EE 144 class.
Documentation for our implementation, as well as discussion of various potential methods to optimize the computation, can be found in the final report here.