Principal Architect — Large Scale Enterprise Solutions, Agentic AI, Security, Privacy and Governance

Adobe · San Jose

onsitefull-time10+ years

posted 12 Aug

ADOBE - Enterprise Customer Experience Platform Principal Architect — Large Scale Enterprise Solutions, Agentic AI, Security, Privacy and Governance San Jose, CA (Hybrid, min. 3-4 days/ week onsite) · Level: Principal · Full-time Summary At A dobe , we are building enterprise-grade Agentic - AI systems that automate large-scale data access and workflow execution for business processes. These systems not only embody foundational enterprise security and privacy considerations but also provide robust and safet y-preserving capabilities in the emerging world of agent identity, delegated access control and human-i n-the-loop execution models. This is a multi -fluency role: you need the distributed systems and Agentic - AI depth to architect how agents reason, plan, and act at scale, and the security-and-compliance mechanisms that manage safe and trustworthy execution. You will partner with engineering, product, legal/privacy, and security leadership to turn fast-moving agentic AI capabilities into a governed, enterprise-ready platform . Key Responsibilities Agentic AI Architecture Define and drive the end-to-end reference architecture for how AI agents are authenticate d and authorized to access enterprise data, tools, and systems across Adobe's Customer Experience Platform — covering agent lifecycle management, interaction protocols, memory/context handling, and action-execution frameworks. Evaluate and select frameworks, protocols, and platforms (e.g., agentic orchestration frameworks, MCP-style tool systems, policy engines such as OPA/Cedar-class authorization) that best support secure, governed multi-agent systems at Adobe's scale. Security, Compliance & Privacy by Design Own the security architecture for agent-to-system and agent-to-agent interactions: identity and workload authentication, least-privilege authorization, secrets and credential mediation, network egress controls, runtime isolation, and telemetry/audit evidence sufficient to satisfy internal audit and external regulators. Embed privacy-by-design and compliance requirements (data governance, consent, data residency, retention) directly into the agent architecture rather than bolting them on after the fact; partner with Legal/Privacy to keep architecture decisions ahead of regulatory expectations for autonomous AI systems. Cross-Organizational Leadership & Standards Lead architectural reviews and set standards for scalability, resiliency, observability, and cost efficiency — with security and compliance treated as first-class, non-negotiable design constraints rather than downstream signoffs . Contribute to or lead an Architecture Review Board function for agentic AI, ensuring new capabilities are evaluated consistently against Adobe's governance bar before they ship. Technical Leadership, Mentorship & Evangelism Serve as Adobe's senior technical authority and encouraging leader on secure agentic AI architecture . Mentor senior engineers and architects. Elevate security and privacy proficiency across the broader engineering organization. Create and maintain architecture documentation, decision records, and threat models that are equally useful to implementation teams and to senior leadership. Executive & Customer Engagement Represent Adobe's agentic AI security and governance posture to executive leadership , enterprise customers, and — where relevant — regulators and standards bodies, with clear, bounded claims about maturity, open risks, and the evidence required before broader rollout. Provide thought leadership across the organization where agentic AI access governance is heading, and shape Adobe's roadmap accordingly. Basic Qualifications 20+ years of experience in software engineering or security architecture, with a strong focus on distributed systems, platform engineering, and enterprise-scale access control. Deep understanding of agentic AI architecture patterns : agent orchestration, tool/function calling, reasoning and planning loops, and multi-agent