Product Management, Human Data Platform

About the Role

Anthropic's Human Data Platform team builds systems designed to collect data that improves our models. This includes the infrastructure to simulate real-world environments and tasks, novel interfaces for data vendors to use, and the pipelines that enable researchers to gather high-quality data at scale. As Claude's real-world usage evolves, so do our data needs — and our tooling has to keep pace. You'll work alongside an engineering team that's quickly prototyping and shipping, help make smart bets about where to focus, and ensure we're investing in tooling that scales. You'll work across research teams, data ops, and external vendors, translating what you learn into clear direction on what to build next.

Responsibilities

• Own the product direction for our human data tooling, with clear prioritization across labeling interfaces, infrastructure investments, data quality, and operational visibility
• Partner with engineering to scope and ship quickly, staying close to the work in a fast-moving prototyping environment
• Develop a deep understanding of research and training approaches to identify where tooling investments will have the highest leverage
• Identify patterns across one-off requests and push toward reusable infrastructure that compounds over time
• Sit in on crowd worker and vendor sessions to systematically understand pain points
• Define and track outcome-based KPIs: time-to-launch for new data collection projects, end-to-end data quality scores, and measurable impact on model evaluation scores

You May Be a Good Fit If You

• Believe that advanced AI systems could have a transformative effect on the world and are interested in helping make sure that transformation goes well
• Are drawn to ambiguous, high-stakes environments where you’ll play a big role in defining the product strategy
• Shipped products where they had to deeply understand technical constraints, not just translate requirements
• Experience working directly with research teams, ideally in AI/ML contexts
• Are e