Software Engineer, Pretraining Data
Core
Build large-scale data systems, pipelines, and infrastructure to turn raw web, code, and multimodal data into training-ready datasets for frontier pretraining runs.
Role type
Senior IC software engineer (distributed systems & data infrastructure)
Builds
Data platform, crawling systems, and data quality pipelines for model pretraining
Domain
AI/ML pretraining data infrastructure
Deliverable
production ML models
Required skills
distributed systems, large-scale data infrastructure, pipeline orchestration, data quality systems, high-throughput processing, system observability, performance-critical coding, experiment design
Preferred skills
staff-level ownership, end-to-end architecture, complex system debugging, AI agent collaboration, intuition for large-scale distributed systems
Technologies
web crawling frameworks, data processing pipelines, telemetry systems, model training infrastructure
Responsibilities
Build and own pipelines that process frontier-scale data with end-to-end traceability; Design and run scaling-ladder experiments on data-mixture and quality depth; Build and scale web crawling systems to discover and parse high-quality documents; Partner with training teams to close the loop on data improvements that move loss and evals; Create clear signals for data quality, lineage, and pipeline health; Debug and harden complex crawl infrastructure for availability and recovery