Assembled

Frontend We're rewriting our calendar component, which was previously built off a heavyweight visualization library, to facilitate a better experience for managing and viewing schedules. The new component will, among other things, efficiently handle a large number of events, allow drag-and-drop and other GCal-esque conveniences, and support zooming in and out by time scale.

Backend We have a mission-critical webhook integration that allows end-users to send us anonymized ticket event data. Our endpoint needs to be highly available—too many dropped events can cause webhook publishers to be disabled automatically. We're building a queue / consumer-based system where downstream breakages won't prevent a successful response and unprocessed events can be easily replayed.

Machine Learning The industry standard for calculating staffing requirements relies on a hundred-year-old formula called Erlang-C. Though quite elegant, it is only tied to real-world data through a handful of parameters. We're building a machine-learning based add-on to ensure its predictions can be compared to and tuned against operationally important metrics like response time, service level, and utilization.