Abstract:  I will present an experimental programming framework, developed at Google, for writing distributed low-latency programs. In this framework, the user expresses their computation as a graph of operators, and the framework takes care of distributing work and communicating across threads and machines.  As such, the framework is targeted as a replacement for there-and-back RPC models widely used at Google and elsewhere.

In my talk, I will go over the details and benefits of the framework, ranging from ease of programming to automatic scaling and optimizations.  The framework is currently deployed in Google Search backends, powering your searches worldwide.

Bio:  Srdjan received his Ph.D. in Computer Science from Dartmouth in 2005, working with Professor Jayanti.  He joined Google right out of school and has been living in a bubble ever since, enjoying coding and free M&Ms.