B

Box

Builds secure content software for businesses

Staff Product Manager - AI Security

Technical Product ManagementConfirmed live in the last 24 hours
Full-timeRedwood City, CA, USA

About the Job

Requirements

  • At least 8 years of product management experience building and scaling enterprise B2B SaaS security products, cybersecurity platforms, cloud security solutions, or AI/ML infrastructure platforms.
  • Deep domain expertise in AI/ML security, large language model application security, prompt injection defense, agentic guardrail frameworks, or model alignment and safety systems.
  • A proven track record developing product strategies around agentic architectures, multi-agent coordination, tool-use and function-calling security, or model integration protocols such as Model Context Protocol.
  • A strong understanding of emerging threat vectors targeting autonomous AI systems, including multi-turn adversarial attacks, indirect injection via ingested content, and coordinated agentic attack swarms.
  • Technical acumen to partner with AI/ML research and engineering leaders on vector embeddings, semantic evaluation models, large language model latency trade-offs, API gateway security, and zero-trust data access control.
  • Ability to define guardrail telemetry, track threat mitigation efficacy, balance safety barriers with responsive AI interactions, influence C-level security leaders, and align cross-functional stakeholders.

Responsibilities

  • Define and execute product strategy, technical vision, and roadmap for AI Security and Agentic Guardrails across Box AI, Box Shield, and the broader Box platform ecosystem.
  • Build agent guardrail frameworks and policy engines to monitor, evaluate, and constrain autonomous agent behaviors across Box Agent, custom workflow agents, and external model integrations.
  • Secure agentic connectivity powered by Model Context Protocol and third-party AI APIs by establishing authentication, context boundary isolation, dynamic permissioning, and secure tool execution.
  • Architect defenses against direct and indirect prompt injection, jailbreaking, training data extraction, system prompt leakage, and coordinated agentic attack swarms.
  • Partner with Engineering, Applied AI/ML Research, Security Architecture, and UX Design to deliver low-latency, enterprise-grade AI security controls while preserving developer velocity and user experience.
  • Drive business and security efficacy metrics, including guardrail evaluation accuracy, prompt injection prevention rates, Model Context Protocol integration adoption, and Shield/Enterprise Plus security revenue growth.
  • Collaborate with Product Marketing, Sales, Customer Success, and Compliance to lead go-to-market strategy and position Box in enterprise agentic AI security.
  • Engage enterprise customers, AI security working groups, and industry analysts to anticipate evolving AI attack vectors, validate product requirements, and guide industry best practices.

Desired Qualifications

  • Experience working with frontier AI/large language models, prompt evaluation frameworks, red-teaming AI systems, or developing real-time guardrail runtime libraries such as NeMo Guardrails, Llama Guard, or proprietary guardrail pipelines.
  • Hands-on familiarity with Model Context Protocol, agent orchestration frameworks, or semantic router architectures.
  • Familiarity with enterprise compliance standards, zero-trust architectures, data loss prevention, or AI governance frameworks such as NIST AI RMF and ISO 42001.
  • A technical background with a Bachelor's or Master's degree in Computer Science, Cybersecurity, Artificial Intelligence, Engineering, or equivalent practical experience.

About the Company

Box

Builds secure content software for businesses

Box builds cloud software for businesses and public agencies to manage documents and other business files, with tools for secure collaboration throughout their lifecycle. Its content platform brings file sharing, document approvals and electronic signatures into one place, so teams work with a shared set of files. Users access content through web, desktop and mobile apps, while administrators control permissions and retention rules. Box AI searches and summarizes business documents, and the company sells subscriptions priced by user alongside premium features and consulting services. Founded in 2005, Box is headquartered in Redwood City, California.

Company size
1,001-5,000
Company stage
IPO
Headquarters
Redwood City, CA
Founded
2005

Recently Posted Jobs

Explore more opportunities at Box

See more jobs