Testing and monitoring the quality, functionality and availability of search and recommendation systems on machine learning

Aleksandr will talk about what hh.ru is doing to not just test, but to build quality into their systems: from the idea and design to functionality, quality and resource-intensiveness/accessibility offline and online. And also, what it takes to ensure that a highly loaded, online-based big data processing system with a black box "Model Stack and Trait Calculation Code" inside has high availability and that its quality and functionality don't degrade.