Research Engineer, Synthetic Data

About this role

Join an early-stage AI infrastructure startup to design and build end-to-end synthetic data pipelines for reinforcement learning environments. You'll transform domain expertise into realistic, scalable training tasks and develop metrics to evaluate their effectiveness for AI agent learning.

What you'll do

• Build and maintain synthetic data pipelines that convert domain workflows into structured training tasks
• Collaborate with subject-matter experts to design high-quality synthetic tasks across professional domains
• Engineer task generation methods that produce diverse, realistic, and learnable data at scale
• Create tooling for mutation, validation, and iterative improvement of synthetic tasks
• Analyze agent performance on synthetic tasks to identify what they teach and failure modes
• Develop metrics quantifying task diversity, realism, learnability, and overall quality

What they're looking for

• Python proficiency
• Docker and Linux environments
• Synthetic data research methods
• End-to-end pipeline design and development
• Environment, evaluation, and benchmark design
• First-principles reasoning about task design and scoring functions
• Data quality assurance and edge case identification
• Async collaboration and technical communication

Benefits

• Competitive salary: $150,000–$250,000 annually
• Equity participation in well-funded early-stage AI company
• Visa sponsorship available
• Work with Olympiad medalists and published researchers
• On-site office environment in San Francisco
• Opportunity to shape foundational AI infrastructure