I am particularly interested in the problem of efficiently collecting high-quality, large, complex, human-annotated datasets for the sake of training machine learning models. While there is a large body of literature on the topic of aggregation and quality control of simple binary or multiple-choice labels, we need to support more complex crowd annotation tasks, including but not limited to: linguistic text annotation such as text span highlighting, syntactic parse trees, semantic parses, translation, paraphrasing, etc; image annotation such as bounding boxes, keypoints, segmentation, captioning; miscellaneous tasks collecting data structures such as ranked lists, clusters, graphs, etc. My current work is on modeling the generative processes by which such data is produced by crowd and expert annotators.