Agentic AI Product Manager — Build & Validate Safe AI Tools

Aisafety; San Francisco (CA); On-site; USD 170,000 - 220,000; Full time

Benefits offered by this job

Health insurance

401K with 4% match

Unlimited PTO

Office meals (lunch & dinner)

Professional development stipend

Referral bonus program

Job summary

The Center for AI Safety (CAIS) is seeking a technical, builder-oriented Product Manager to own an agentic product. You’ll quickly learn the problem space, define the roadmap, and drive execution end-to-end with the underlying agentic architecture and AI coding tools.

You’ll prototype with Claude Code or similar, collaborate with researchers and operations, run user research with technical and non-technical stakeholders, and own metrics and outcomes while embracing ambiguity.

Qualifications

• 3–5 years of Product Management experience on technical products or AI features.
• Hands-on coding or engineering experience with shipped code.
• Fluency with AI coding tools (Claude Code, Cursor, GitHub Copilot).
• Strong technical fluency: SQL, Python, APIs, notebooks, or Git.
• Track record of writing clear PRDs and shipping products 0→1 or scaling.
• Comfortable operating in ambiguity and fast-paced environments.

Responsibilities

• Take ownership of an agentic product need, building from scratch or accelerating existing work.
• Define the go-forward product roadmap and prioritize features, managing scope and tradeoffs.
• Drive roadmap execution and iterate based on user feedback, usage data, and testing.
• Prototype and validate changes using AI-assisted coding tools.
• Partner cross-functionally with researchers, operations, and external partners to ship quickly.
• Conduct user research with technical and non-technical stakeholders.
• Own metrics and outcomes; use experiments to validate decisions.
• Represent the user through demos, briefings, and customer conversations.

Skills

PM experience in technical products

Hands-on coding

AI coding tools

SQL

Python

APIs

PRD writing

Communication

Ambiguity tolerance

Tools

Claude Code

Cursor

GitHub Copilot

Job description

The Center for AI Safety (CAIS) is seeking a technical, builder-oriented Product Manager to own an agentic product. You’ll quickly learn the problem space, define the roadmap, and drive execution end-to-end with the underlying agentic architecture and AI coding tools.

You’ll prototype with Claude Code or similar, collaborate with researchers and operations, run user research with technical and non-technical stakeholders, and own metrics and outcomes while embracing ambiguity.