When collecting multi-modal human data, establishing a reliable "ground truth" can be notoriously difficult and subjective. We are looking for an ML Data Researcher who will tackle this exact challenge. Your goal will be to design and conduct experiments on our data collection methods to discover techniques that reduce subjectivity, improve signal quality, and ultimately help our multi-modal models converge faster and achieve higher performance.
BS in Computer Science, Machine Learning, Statistics, Neuroscience, or a closely related field
Hands-on experience in a data-oriented, experimentation-heavy role, ideally supporting ML model development
MS/PhD in Computer Science, Machine Learning, Statistics, Neuroscience, or a closely related field
10 years of relevant industry experience
Experience with multi-modal human data (e.g., video, audio, motion capture) and the challenges of establishing ground truth for such data
Familiarity with human probing/annotation methods and inter-rater reliability techniques
Experience designing data collection experiments and interpreting their effect on model training/convergence
Outstanding communicator with the ability to present complex technical findings to diverse audiences
Collaborative team player who thrives in multidisciplinary, cross-functional environment