Research Engineer - Pretraining Data

Role overview

Build and improve the datasets used to train the next generation of open AI models. The work spans discovery of useful web, code, and technical material; scalable preparation of raw sources; and collaboration with researchers to assess how data affects broad pretraining and capabilities such as reasoning, coding, and long-context work.

Responsibilities - Find, assess, and collect useful training material from web, code, and technical sources. - Build scalable pipelines to extract, clean, deduplicate, and filter raw data into training-ready corpora. - Create checks for data quality, coverage, and contamination. - Work with researchers to evaluate which sources and data treatments improve model performance. - Contribute across data acquisition, scalable processing, or data research according to your area of expertise.

Requirements

The available source does not specify a complete list of required qualifications. It indicates that experience in data acquisition, scalable processing, or data research can be a relevant starting point, and that candidates need not cover every area.

Benefits and work setup - Full-time role with remote work or in-person work in San Francisco. - Stated cash compensation range of $150,000–$350,000, including equity incentives. - Visa sponsorship and relocation assistance are available for international candidates. - Quarterly team off-sites, hackathons, conferences, and learning opportunities.