We build a probabilistic graph of internet-connected devices based on billions of signals. Lots of Scala-based big data problems to hack on! The realtime systems handle many 100s of thousands of QPS with mere millisecond latency.
How can you detect and remove anomalous data (from botnets perhaps) from datasets that are multiple petabytes in size? Can you write a connected components algorithm that works efficiently at such a scale? Can you write a model that detects individual behaviors within a cluster of device activity?
We run our own datacenters globally, are migrating systems over to Mesos (ooh, shiny), and infrastructure is a first-class critical project, not an afterthought.
All of this happens with a fairly small team of just over 30 engineers, all working together in NYC.
Come join us.
Email toby@tapad.com and say hi.