Work on news feed integrity modeling team.
Main project consists of improving labeling quality by creating new labeling queues. This includes creating new labeling screens that both showed content of sample and decision widget. Further, project involves designing data pipeline to aggregate results in the backend so that these could be incorporated into training dataset.
Side projects include experimenting with in-house APIs to incorporate new signals into our models and analyzing impact of these on model's precision and recall statistics, and enabling rendering of new video features on labeling screens.