Software Engineer, Pretraining Data

What you'll do

• Build and operate high-throughput data pipelines, orchestration, and tooling for reliable, observable, reproducible pretraining-data iteration.
• Train and deploy models to classify, rank, filter, and clean data, and run experiments to evaluate data quality and its effects on model performance.
• Develop web crawling and parsing infrastructure that discovers, fetches, and processes public web content for model training.

What they ask for

• Strong infrastructure or data platform background.
• Ability to architect and ship end-to-end, debug complex systems independently, and work alongside AI agents.
• Strong intuitions about large-scale distributed systems.

Nice to have

• Exceptional depth in a domain such as crawling or search infrastructure.

AI focus

Builds data systems and trains models that prepare and evaluate data for frontier model pretraining; the role also involves working alongside AI agents.