Senior Enterprise AI Security Engineer

Box•Redwood City, CA

22h•$190,000 - $263,000•Hybrid

About The Position

Box has a world-class security organization responsible for protecting our customer and employee data around the globe. We are looking for a Senior Enterprise AI Security Engineer to join our Enterprise Security team and own how Box adopts AI internally — safely, quickly, and at scale. This role pairs enterprise security architecture with hands-on engineering. You will assess the AI tools and agent workflows Boxers want to use, then convert each assessment into a control that holds for everyone who adopts something similar later. You do not need to have trained a model, but you do need working intuition for the failure modes: an agent handed more authority than its task requires, untrusted text arriving as instructions, or company data leaving quietly inside an API call. You will partner closely with IT, Infrastructure, Incident Response, Legal, Privacy, and the business functions piloting AI, shaping the technical direction of Box's secure AI adoption program.

Requirements

Bachelor's or Master's degree in computer science, information security, or a related field, or equivalent practical experience
6+ years in enterprise or corporate security engineering, security platform engineering, application or product security, or a combination — with meaningful recent time spent on AI or agentic systems
Hands-on experience securing real AI deployments — enterprise assistants, internal agents and the tool ecosystems around them, grounded search or RAG over company data, or AI coding tools inside a developer environment
A working mental model of how AI systems get abused and paired with the ability to analyze AI workflows to determine guardrails.
Strong identity and access management fundamentals
Experience with security automation in Python, Go, or a similar language, and a track record of building tooling, libraries, or platform controls
Fluency in the emerging problems of this space: shadow AI, agentic and non-human identity, SaaS-to-SaaS integration risk, browser security, and data exfiltration paths
Exceptional communication skills, with the ability to explain AI risk and its business impact to practitioners
Proven ability to work independently with autonomy, manage ambiguity, and recommend secure but risk-profiled decisions

Nice To Haves

A track record of shipping internal security platforms
Exposure to isolation and permissioning for autonomous workloads, or to AI red teaming and evaluation work

Responsibilities

Design, build, deploy, and maintain comprehensive security systems that allow Box's workforce use AI tools and agents productively without losing visibility, identity boundaries, or control of company data.
Architect and operate the control plane for workforce AI: AI gateways, prompt and response logging, and egress controls across approved assistants and copilots/agents
Extend identity and access management to agentic and non-human identities — scoping, credential lifecycle, delegation, OAuth grant review, and revocation for agents and connectors reaching into corporate SaaS
Gate what internal agents can reach: an approval path for MCP servers and connectors, scoped permissions per tool, and isolated execution for anything that runs code
Harden the AI-era endpoint and browser layer: enterprise browser policy, extension governance, AI-assisted developer tooling, and local agent runtimes
Build the shared enforcement layer — policy engines, brokers, approval registries, internal libraries — that catches risky behavior in flight: injected instructions, attempts to talk around a restriction, tools used outside their intended purpose, sensitive data heading somewhere it should not, and agents behaving nothing like they did last week
Lead security architecture reviews and threat modeling for AI vendors, AI features enabled inside existing SaaS platforms, and internal agent deployments, tracing where untrusted content, company data, and privileged credentials meet in each one
Own the standards, reference architectures, and paved paths that define what approved AI usage concretely means at Box, and build reusable patterns
Partner with IT, Infrastructure, Incident Response, Engineering, Legal, and Privacy to mature AI governance, including AI-specific response playbooks
Automate through scripting, infrastructure-as-code, and orchestration instead of manual review queues, and bring AI-assisted investigation tooling to the security team itself
Help raise the technical bar across the team through design and code review, mentoring and guidance