Re/rs, Data Understanding - Foundations

OpenAI OpenAI · AI Frontier · San Francisco, CA · Research

This role focuses on research and development of high-quality datasets for large model training at OpenAI. Responsibilities include synthesizing data, building VQ representations, and processing/filtering data. The role involves treating data quality as a research problem, developing new methods for data selection and transformation, and designing experiments to understand data's impact on model learning. The goal is to translate research into scalable data processing pipelines.

What you'd actually do

  1. creating the high quality datasets and their quantized representation for OpenAI.
  2. synthesizing data, building VQ representations, and processing, filtering, deduplication, quality control, and tokenization so it can be used effectively in big model training runs.
  3. treat data quality and curation as core research problems: developing new methods to select, combine, and transform data; creating datasets that improve model capabilities; and designing rigorous experiments to understand how data choices and interventions affect model learning and downstream behavior.
  4. work closely with frontier models and web-scale data to build evidence for which approaches work and why, then translate successful research into scalable data processing pipelines
  5. Have a strong track record of new or improved ML ideas, through publications, projects, or applied research.
  6. Own and drive a research agenda, from choosing the right problems to carrying long-running work through to impact.

Skills

Required

  • ML ideas
  • publications
  • projects
  • applied research
  • research agenda
  • data quality
  • data curation
  • data selection
  • data transformation
  • model capabilities
  • model learning
  • downstream behavior
  • frontier models
  • web-scale data
  • scalable data processing pipelines

Nice to have

  • AI’s impact, including privacy, provenance, and data quality
  • building high-performance deep learning systems
  • large-scale data processing systems

What the JD emphasized

  • strong track record of new or improved ML ideas, through publications, projects, or applied research
  • Own and drive a research agenda

Other signals

  • synthesizing data
  • building VQ representations
  • processing, filtering, deduplication, quality control, and tokenization
  • big model training runs
  • data quality and curation as core research problems
  • developing new methods to select, combine, and transform data
  • creating datasets that improve model capabilities
  • designing rigorous experiments to understand how data choices and interventions affect model learning and downstream behavior
  • work closely with frontier models and web-scale data
  • build evidence for which approaches work and why
  • translate successful research into scalable data processing pipelines