Research Engineer, Synthetic Data

About this role

Build synthetic data pipelines that transform domain workflows into scalable AI training tasks. Work with a selective team of researchers and engineers to design generation methods, validation systems, and quality metrics that enable AI agents to learn complex behaviors.

What you'll do

• Design and implement end-to-end synthetic data pipelines converting domain workflows into structured training tasks
• Collaborate with subject-matter experts to generate realistic synthetic tasks across professional and technical domains
• Develop scalable task generation methods producing diverse, realistic, and learnable outputs
• Build automated tooling for task mutation, validation, and continuous quality improvement
• Analyze model and agent performance on synthetic tasks to identify learning patterns and failure modes
• Define and implement metrics for task diversity, realism, learnability, and overall data quality

What they're looking for

• Python and Linux development
• Docker and containerization
• ML infrastructure and data pipeline design
• Synthetic data systems and methodology
• Evaluation frameworks and benchmarking for AI agents
• Reinforcement learning and agentic AI workflows
• LLM post-training pipelines
• First-principles reasoning and edge-case detection

Benefits

• Visa sponsorship available
• Competitive salary range $150,000–$250,000
• Work with Olympiad medalists and published researchers
• Direct ownership over generation methods and validation systems
• On-site collaboration in San Francisco (Singapore candidates considered)