Product Manager, Claude Code Model Performance

About the role

Anthropic is looking for a Product Manager to own the end‑to‑end launch and performance evaluation of Claude Code, the world’s most capable coding agent. You will bridge frontier AI research with the daily needs of millions of developers, turning research breakthroughs into shipped improvements.

Key responsibilities

• Own model launch planning and execution: define readiness criteria, coordinate research and engineering, and ensure clean releases for developers.
• Design and implement agentic evaluations that measure real‑world coding performance.
• Drive the engineering team’s evaluation roadmap.
• Partner with researchers to define target behaviors and influence model development using real‑usage evidence.
• Engage with users, analyze transcripts, and translate capability gaps into product improvements.
• Synthesize signals from internal users, external developers, and competitive benchmarks into clear priorities.

Required profile

• Built agentic evaluations (e.g., SWE‑bench‑style task suites).
• Daily user of Claude Code with a clear sense of desired model behaviors.
• Engineering background with 2+ years in product management or equivalent product‑direction experience.
• Deep understanding of AI concepts, model behavior, prompt engineering, and evaluation methodology.
• Systems thinker who builds infrastructure to prevent recurring problems.
• Experience launching products or capabilities in ambiguous, research‑adjacent environments.
• Creative, hacker spirit with a passion for solving puzzles.

Required skills

• Agentic evaluation design (e.g., SWE‑bench).
• Prompt engineering.
• AI model evaluation methodology.