Product Manager, AgentWarden (Agent Security)
Hybrid (San Francisco, CA)$125k - $165kFull-timePosted 2w ago
About the role
Product Manager for AgentWarden, Dynamo AI’s agent security product. Own the problem of getting AI agents approved and operating safely in production, working with engineering, research and enterprise security customers.
Responsibilities
- Own the AgentWarden roadmap against a defined set of customer problems and success metrics.
- Lead customer discovery and pilot engagements with enterprise security and AI platform teams, and present to senior security stakeholders.
- For each deployment, identify the buyer, the approver and the operator, and shape pilots around what each of them needs in order to proceed.
- Write requirements that engineering and research can build and evaluate against, including test sets and acceptance criteria.
- Track agent frameworks, MCP, agent platforms and related threat research, and update the roadmap as they change.
- Work with the DynamoGuard and DynamoEval PMs on shared policies, evaluations and alerting.
- Pre-deployment risk assessment for agents: analysis of agent tools, permissions and runtime behavior, and the evidence security teams need to approve an agent.
- Red teaming for agents, covering multi-step attacks, tool misuse and indirect prompt injection.
- Runtime controls for agent actions and tool calls, kept current as new attack techniques are published.
- Alerting and investigation for agent incidents, where a single session can span many steps and systems.
Qualifications
- 2 to 4 years of product management experience, preferably in security, developer tools, or AI/ML products sold to enterprises.
- Hands-on familiarity with how agents are built, including tool calling, MCP, and at least one agent framework such as LangGraph, the OpenAI Agents SDK or CrewAI. You should be able to build a simple agent yourself.
- Working knowledge of LLM and agent threats, such as prompt injection, excessive agency and data exfiltration, and of references such as the OWASP Top 10 for LLM Applications.
- Ability to read Python or TypeScript well enough to review scanner findings and discuss detection approaches with engineers.
- Experience running customer discovery with technical buyers and converting it into scoped requirements.
- Clear written and verbal communication, including presenting to enterprise security leaders.
- Willingness to spend 4 to 6 weeks per year at customer sites, and more during large contract engagements.
Skills
Required:Product managementAgent/tooling familiaritySecurity and threat knowledgeCustomer discovery and storytelling
Nice to have:Python/TypeScript reading
Last updated Oct 7, 2026. Interested? Apply on the company site.
About Dynamo AI
Compliant-Ready AI for the Enterprise
End-to-end privacy, security, and compliance solutions to prepare your organization for emerging AI regulations.
machine-learningprivacydata-engineering
More open roles at Dynamo AI
- Product Manager, AI Governance & Compliance2w agoHybrid (San Francisco, CA)
- Product Analyst3mo agoHybrid (San Francisco, CA) · Product
- ML Research Scientist Intern – Dynamo Guard / Dynamo Eval / AgentWarden8mo agoSan Francisco, CA, New York, NY +2 more · Engineering
- ML Research Engineer Intern – Dynamo Guard / Dynamo Eval / AgentWarden8mo agoHybrid (New York, NY; San Francisco, CA; Remote)
- Senior Software Engineer9mo agoHybrid (Remote (US)) · Engineering