I'm partnering with a ~$9B frontier AI company building next-generation embodied AI and world models.
They're looking for an exceptional Principal/Staff+ Research Engineer to tackle one of the hardest problems in frontier AI:
Given massive amounts of multimodal data and compute, how do you determine which data will actually make the next model better - before you train it?
You'll work across:
Data quality, valuation, selection & curation
Massive-scale image, video & multimodal datasets
Dataset enrichment and evaluation
Understanding how training data impacts downstream model performance
World models / frontier multimodal AI
Technical strategy across research, data & model teams
This is a highly senior IC role with significant technical influence across the organization, not simply a data pipeline or infrastructure position.
Ideal background:
Experience working with multimodal, image, video, robotics or other large-scale ML datasets
Deep understanding of data quality and its relationship to model performance
Experience operating at genuine frontier-model / production scale
Strong cross-functional technical leadership and judgment
Comfortable solving ambiguous, open-ended research and engineering problems
Autonomous driving experience is NOT required. We're particularly interested in people from frontier AI labs, multimodal/video foundation-model teams, robotics, and other organizations solving data problems at enormous scale.
📍 Bay Area / London + flexibility for exceptional candidates
