Pretraining Data Engineer - Scalable Pipelines

Cursor; San Francisco (CA); On-site; USD 180,000 - 260,000; Full time

Job summary

Cursor seeks software engineers to build the data systems behind frontier AI model training. You will work on crawlers, data pipelines, and training-ready datasets, turning raw data into tokens and mixes for model pretraining.

You will collaborate with Data Quality, Data Platform, and Crawling teams to ensure data is accurate, timely, and scalable. You’ll own end-to-end data pipelines, improve throughput, and partner with researchers to turn data ideas into measurable improvements in loss and

Qualifications

• Experience with large-scale data infrastructure and pipelines.
• Ability to diagnose and optimize data quality and throughput.
• Demonstrated ownership of complex systems from design to production.

Responsibilities

• Build and scale data pipelines for frontier-scale data.
• Own pipelines, orchestration, and tooling for fast pretraining iterations.
• Signal data quality and lineage to researchers.
• Collaborate with data acquisition and training teams to improve loss and evals.
• Develop performance-critical code and experiments.

Skills

Distributed systems

Data pipelines

Crawling infrastructure

End-to-end ownership

Tools

Airflow

Kafka

Spark

Job description

Cursor seeks software engineers to build the data systems behind frontier AI model training. You will work on crawlers, data pipelines, and training-ready datasets, turning raw data into tokens and mixes for model pretraining.

You will collaborate with Data Quality, Data Platform, and Crawling teams to ensure data is accurate, timely, and scalable. You’ll own end-to-end data pipelines, improve throughput, and partner with researchers to turn data ideas into measurable improvements in loss and