We are designing concurrent and parallel systems that greatly improve performance and significantly reduce power consumption, thereby serving as the basis for the upcoming era of massive data processing. Our general strategy in attacking this goal is (1) to holistically understand computer systems from microarchitectures to programming languages algorithms, and (2) to develop abstraction layers that realize the intrinsic parallelism of the workloads while providing easy programming environment at the same time. Specifically, we are working on the following projects: