Research Product Manager, Model Behaviors

Anthropic•San Francisco, CA

12h•$385,000 - $460,000•Hybrid

About The Position

As a Product Manager for Model Behaviors, you will partner with the Alignment Finetuning team to define and shape Claude's character, behaviors, and reinforcement signals—work that directly influences how millions of people experience AI. You will systematically identify high-priority behavioral improvements, coordinate across Research, Product, and Safeguards teams, and accelerate our ability to ship well-aligned models. The ideal candidate combines deep user empathy with the judgment to navigate nuanced behavior questions where there are no clear right answers.

Requirements

Have a deep passion and curiosity for AI and LLMs. Use AI regularly.
Have 5+ years in product management leading scaled conversational AI products.
Are a first-principles thinker with the ability to navigate and execute amidst ambiguity, flexing into different domains based on the business problem at hand and finding simple, easy-to-understand solutions
Have a track record of delivering products and features to end-users (consumer or end-user b2b focus)
Have strong user empathy and the ability to synthesize vague or contradictory feedback into actionable priorities
Have strong judgment and model taste, with the ability to make tradeoffs when there is no clear right answer
Have a strong grasp of ML concepts and are willing to go deep on technical solutions
Have intellectual curiosity without ego—comfortable asking questions and learning independently
Think creatively about the risks and benefits of new technologies, moving beyond past checklists and playbooks
Have a creative, hacker spirit and love solving puzzles

Responsibilities

Define behavioral defaults and steerability constraints
Develop and maintain taxonomies of model behaviors across capabilities
Identify, triage, and prioritize behavior issues and opportunities, coordinating input from Users, Research, Product, and Safeguards teams
Amplify alignment research breakthroughs, translating them into product, process, and model improvements
Deeply understand user interaction patterns to identify behavior improvements that make Claude more helpful and safe
Contribute to evals that measure alignment progress
Identify and scale initiatives and tools that help researchers ship alignment improvements faster