Frontier AI Data Pipelines Lead

Greylock Partners; San Francisco (CA); Hybrid; USD 110,000 - 160,000; Full time

Job summary

Greylock Partners is seeking a hybrid role at the intersection of scientific domain expertise, AI, and operational rigor. You will own the full pipeline that turns real scientific workflows into training data for frontier AI models, recruiting elite researchers to generate high-quality examples of those tasks, and running the operational machine that delivers the dataset to frontier labs.

The ideal candidate is a scientist or technically trained operator who has published research, understands

Qualifications

• A science-related degree from a top university.
• Experience producing and publishing research papers.
• 2+ years in an operationally rigorous role (e.g., consulting, private equity, or domain-specific fields).
• High comfort with AI tools and rapid decision-making in ambiguous settings.

Responsibilities

• Own data labeling programs end to end, turning scientific workflows into training signals for AI models.
• Partner with researchers and domain experts to set the quality bar for scientific reasoning.
• Design and oversee end-to-end data pipelines from customer specs to final delivery.
• Identify bottlenecks, restructure workflows, and improve quality, throughput, and cost-efficiency.
• Build an elite network of scientific contributors and lead them to create frontier datasets.

Skills

Science background

Research publishing

Operations

Ambiguity tolerance

Education

Science-related degree from a top university

Job description

Greylock Partners is seeking a hybrid role at the intersection of scientific domain expertise, AI, and operational rigor. You will own the full pipeline that turns real scientific workflows into training data for frontier AI models, recruiting elite researchers to generate high-quality examples of those tasks, and running the operational machine that delivers the dataset to frontier labs.

The ideal candidate is a scientist or technically trained operator who has published research, understands