My research focuses on data-centric machine learning, emphasizing the often-overlooked aspects of data such as sourcing, annotation, and validation, which critically impact the reliability and usability of ML systems. I combine theoretical and experimental methods to develop conceptual insights with practical relevance.

Within data-centric machine learning, I am especially interested in:

• Active Learning and Experimental Design (methods to select data to collect supervision)

• Statistical aspects of data algorithms and data-centric ML (e.g., noise, domains, concept shift)

• Task specifications with instructions (e.g., prompts, concepts) and demonstrations (e.g., labeled examples)