Product Manager, Claude Science
San Francisco, CA | New York City, NY | Seattle, WA
Posted 14d ago
About the job
Anthropic's mission is to create reliable, interpretable, and steerable AI systems, believing that powerful AI can accelerate scientific progress. Claude Science is an AI workbench designed to provide researchers with a unified environment for tasks that currently span numerous disconnected tools, including literature review, specialized database access, scientific computing, analysis, and publication preparation. It already supports native rendering of protein structures, genome tracks, and chemical structures, and coordinates multi-agent workflows with built-in error checking for citations and calculations. The team is expanding the Claude Science product group, seeking individuals to own significant roadmap areas end-to-end, such as expanding into new scientific fields, integrating new model capabilities with researchers, or preparing the workbench for enterprise R&D. This role involves direct engagement with researchers in academic labs, biotech, and pharma R&D to understand scientific workflows from hypothesis to publication, translating these insights into product features. Collaboration with research teams will focus on enhancing the models' scientific capabilities, identifying target fields and workflows, and ensuring Claude Science is trusted in enterprise and regulated research settings.
Responsibilities
- Own vision, strategy, roadmap, and execution for a major area of Claude Science, and contribute to overall product direction in a nascent category.
- Engage deeply with working scientists through interviews and working sessions to synthesize learnings into clear priorities, requirements, and success metrics.
- Partner with research to improve Claude's scientific capabilities by defining target behaviors, building and shaping evaluations grounded in real scientific workflows, identifying usage-based failure modes, and feeding this information back into model development.
- Drive end-to-end launches of models and features by defining readiness criteria, coordinating across research, engineering, design, and go-to-market teams, managing early access programs, and ensuring successful researcher adoption.
- Identify and prioritize scientific domains, data sources, tools, and integrations for Claude Science to support, building a case with user evidence, usage data, and market analysis.
- Prototype ideas using Claude to validate them before committing engineering resources.
- Collaborate with safeguards, policy, security, and go-to-market teams to ensure responsible deployment of powerful scientific capabilities to trusted researchers, including in enterprise and regulated environments.
- Drive enterprise readiness by addressing security and compliance reviews, admin controls, and procurement realities for large organizations in pharma and biotech.
- Ruthlessly prioritize across scientific domains, customer segments, and workflows, making clear decisions on product focus and MVP versus ideal state.
- Define and utilize metrics for AI research and development products, such as adoption, retention, time-to-result, and trust in outputs, to guide planning.
- Maintain an objective and current understanding of the AI-for-science ecosystem, including models, tools, benchmarks, and competitors, and Claude Science's position within it.
Requirements
- Product management experience shipping technical products in close collaboration with engineering and design, or equivalent experience as a founder, engineer, or scientist driving product direction.
- Scientific or deeply technical background sufficient to engage with domain experts on their workflows and diagnose model performance on scientific tasks.
- Strong understanding of AI and LLM concepts, daily use of AI tools, and comfort with in-depth discussions on model behavior, prompting, and evaluation.
- Data-driven approach to prioritization, grounding decisions in usage data, evaluation results, and user evidence.
- Strong user empathy with the ability to synthesize vague or contradictory feedback from expert users into actionable priorities.
- Ability to navigate and execute in ambiguous environments as a first-principles thinker, demonstrating flexibility across scientific domains.