企业上下文架构师
Enterprise Context Architect
职位描述
Dropbox 正在构建连接内容、上下文和行动的知识层。随着人工智能从助手演进为能够执行任务的系统，企业知识的结构和管理成为决定 AI 是助力还是失败的关键。
该职位将定义 AI 依赖的上下文层的架构和技术需求：AI 能够了解什么、信任什么以及被允许执行什么。你将制定标准和方法，使企业知识变得可靠、及时且具备权限意识，跨企业开展内容和上下文架构、检索与依据、评估、语义结构、权限要求和原型设计工作，同时领域专家仍对内容的准确性和管理负责。
这是 Dropbox 首个此类职位。你将与 IT、工程、法律、隐私和安全团队合作，你的决策将直接影响公司内 AI 的表现。
职责
- 定义企业知识的权威源策略：哪些系统是权威的，哪些内容集中索引而非实时获取，哪些内容适用于 AI 使用，哪些内容被归档或排除，基于对我们高价值流程背后权威来源的评估。
- 定义使内容适合 AI 的企业标准，涵盖结构、元数据、来源和访问权限，包括何时需要语义模型或知识图谱，何时不需要，并将其转化为各领域采用的创作模式。
- 设计 AI 行动的控制模型，包括资格规则、前提条件、审批边界、升级路径和回滚要求，确保基于企业知识执行任务的系统在 AI 能力演进过程中保持可追溯和安全。
- 领导内容堆栈中的平台和连接器策略。推动关于重构、迁移、就地索引或整合的决策，并与 IT 和工程团队合作，确定连接器架构以及 AI 系统如何获得对工具和来源的访问权限。
- 构建企业内容的联邦运营模型：跨职能的管理职责，各领域在共享标准内对其自身准确性负责，以及覆盖审查频率、过期、重大变更触发和退役的生命周期政策，与业务关键性相关联。
- 定义 AI 环境下的内容质量。建立检索和依据评估体系
查看英文原文
Role Description
Dropbox is building the knowledge layer that connects content, context, and action. As AI moves from assistants to systems that act, the structure and stewardship of enterprise knowledge becomes the difference between AI that helps and AI that fails.
This role defines the architecture and technical requirements for the context layer our AI depends on: what it can know, what it can trust, and what it is allowed to act on. You will shape the standards and approaches that make enterprise knowledge reliable, current, and permissions-aware, working across the enterprise on content and context architecture, retrieval and grounding, evaluation, semantic structure, permissions requirements, and prototyping, while domain experts remain accountable for the accuracy and stewardship of their content.
This is the first role of its kind at Dropbox. You will partner with IT, Engineering, Legal, Privacy, and Security, and your decisions will show up directly in how AI performs across the company.
Responsibilities
- Define the source-of-truth strategy for enterprise knowledge: which systems are authoritative, what is indexed centrally versus fetched live, what is eligible for AI use, and what is archived or excluded, informed by an assessment of the authoritative sources behind our highest-value workflows.
- Define the enterprise standards that make content AI-ready across structure, metadata, provenance, and access, including where semantic models or knowledge graphs are warranted and where they are not, and translate them into authoring patterns adopted across domains.
- Design the control model for AI actions, including eligibility rules, preconditions, approval boundaries, escalation paths, and rollback requirements, so systems that act on enterprise knowledge stay traceable and safe as AI capabilities evolve.
- Lead platform and connector strategy across the content stack. Drive decisions on what is refactored, migrated, indexed in place, or consolidated, and partner with IT and Engineering on connector architecture and how AI systems are granted access to tools and sources.
- Build the federated operating model for enterprise content: stewardship across functions, domains accountable for their own accuracy within shared standards, and lifecycle policies covering review cadence, expiration, material-change triggers, and retirement, tied to business criticality.
- Define content quality in an AI context. Stand up retrieval and grounding evaluations for priority use cases, extend measurement to workflow traces and policy conformance as systems begin to act, and route findings back into the content lifecycle.
- Co-own the criteria for AI content eligibility, sensitivity classification, and permissions modeling with Legal, Privacy, and Security, including access boundaries for the tools AI systems can reach.
Requirements
- 7+ years designing how information is structured, governed, owned, and maintained at enterprise scale, including at least 2 years supporting AI-enabled knowledge or retrieval systems.
- Direct experience preparing content for AI consumption, with working fluency in RAG, grounding, semantic chunking, embeddings, vector search, and citations.
- Hands-on experience with knowledge graphs, ontologies, or semantic models for machine-readable content.
- Practical experience evaluating retrieval and grounding quality, testing hypotheses, and building lightweight prototypes independently or with Engineering.
- Track record building federated operating models across functions outside direct reporting lines, including metadata standards or authoring frameworks adopted at scale.
- Demonstrated ability to influence senior stakeholders across Engineering, IT, Legal, Security, and business functions.
- Sound judgment in balancing centralized standards with domain-specific expertise and ownership.
Preferred Qualifications
- Hands-on experience with enterprise platforms such as ServiceNow, Atlassian, Microsoft 365 or Copilot Search, Slack, or Notion.
- Familiarity with structured authoring (such as DITA), controlled vocabularies, or knowledge operations methodologies such as KCS.
- Experience with AI evaluation tooling and frameworks for measuring retrieval quality, groundedness, and answer relevance.
- Background working in regulated, policy-heavy, or high-risk content domains.
- Working knowledge of NIST AI RMF, OWASP GenAI guidance, or comparable risk frameworks.
Durable Skills
AI fluency means using these tools to amplify human judgment, not replace it. We believe people with these skills will thrive as work and technology continue to evolve:
- Awareness: Understand yourself and others.
- Judgment: Evaluate information and make decisions in complex situations.
- Adaptability: Learn, adjust, and stay effective through change.
- Connection: Communicate, collaborate, and build trust.
To learn more about why these skills matter and what the data shows about thriving through change, read this blog post from our Chief People Officer, Melanie Rosenwasser.
Compensation
US Zone 1
This role is not available in Zone 1
US Zone 2
$159,100—$215,300 USD
US Zone 3
$141,400—$191,400 USD