Principal Architect

Overview Security represents one of the most critical priorities for our customers in a world of evolving digital threats, increasing regulatory requirements, growing estate complexity, and rapid advances in AI. Microsoft Security aspires to make the world a safer place by empowering every user, customer, and developer with integrated security solutions that protect across heterogeneous environments.

Perception is an advanced Agentic AI platform designed to empower security and IT teams to operate at the speed and scale of AI while adhering to Microsoft’s responsible AI principles.

We are looking for a Principal Architect to define and drive the architectural vision and long-term technical strategy for the Perception Platform, with a particular focus on enabling AI platform capabilities to operate securely, reliably, and at scale across multiple tenants and diverse enterprise environments .

In this role, you will provide architectural leadership across complex and interconnected areas spanning Generative AI, Retrieval-Augmented Generation (RAG), memory, reasoning, agentic and multi-agent systems, identity and authorization, AI safety, distributed cloud systems, and platform infrastructure .

You will define reference architectures, architectural principles, system boundaries, integration patterns, and reusable platform abstractions that enable sophisticated AI capabilities to support different tenant configurations, infrastructure dependencies, security boundaries, identity and authorization requirements, compliance constraints, and deployment models.

You will drive architectural strategy for reducing tenant-specific coupling and enabling capabilities to be deployed, configured, validated, monitored, and operated consistently across enterprise environments.

This includes guiding the evolution of existing production architectures while establishing common architectural patterns for identity, compute, storage, networking, configuration, deployment, observability, reliability, security, data, and AI execution.

This role requires deep expertise in distributed systems and cloud architecture, strong judgment across complex architectural tradeoffs, a deep understanding of modern AI systems, and the ability to influence technical direction across organizational boundaries.

You will partner with engineering leaders, product managers, architects, security experts, privacy and compliance teams, and partner organizations to translate complex customer, product, security, and platform requirements into scalable architectures and long-term technology investments.

As a Principal Architect, you will be expected to identify systemic architectural challenges, establish long-term technical direction, drive architectural convergence across teams and services, define durable architectural principles and reference patterns, and create technical leverage that extends beyond any individual product, component, or organization .

You will serve as a technical authority for critical architectural decisions, helping teams navigate tradeoffs involving security, reliability, scalability, extensibility, operational complexity, customer requirements, and long-term platform evolution.

This is an opportunity to solve some of the most challenging architecture problems at the intersection of AI, security, distributed systems, identity, and multi-tenant cloud architecture , while helping establish the technical foundation for delivering Microsoft Security AI capabilities securely and consistently to customers at scale.

Microsoft’s mission is to empower every person and every organization on the planet to achieve more. As employees we come together with a growth mindset, innovate to empower others, and collaborate to realize our shared goals. Each day we build on our values of respect, integrity, and accountability to create a culture of inclusion where everyone can thrive at work and beyond.

Responsibilities

• Define the architectural vision and long-term technical strategy for the Perception Platform, establishing principles and reference architectures for production-grade AI capabilities spanning generative AI, retrieval, memory, reasoning, agent orchestration, model interaction, execution, evaluation, extensibility, security, and safety.
• Define scalable multi-tenant reference architectures that enable platform capabilities to operate securely and reliably across multiple tenants and enterprise environments, establishing clear patterns for tenant isolation, identity boundaries, authorization, configuration, data protection, deployment, sovereignty, compliance, and operations.
• Establish architectural principles, system boundaries, and reusable platform abstractions across compute, storage, networking, identity, configuration, deployment, observability, data, AI execution, and service integration that reduce architectural fragmentation and tenant-specific coupling.
• Drive architectural convergence across distributed systems and engineering teams , identifying divergent implementations and establishing common patterns, interfaces, contracts, and platform capabilities where standardization creates meaningful engineering and operational leverage.
• Define an automation-first architecture for tenant provisioning, configuration, deployment, validation, upgrades, migrations, lifecycle management, and operational readiness, enabling consistent operation across heterogeneous tenant environments.
• Define architectural approaches that accommodate variations in tenant configuration, service availability, infrastructure topology, networking, identity, security, data residency, sovereignty, and compliance requirements without propagating environment-specific assumptions throughout the application architecture.
• Establish architectural strategy for scalable Retrieval-Augmented Generation (RAG) systems, including ingestion, indexing, embeddings, semantic and hybrid retrieval, context construction, grounding, provenance, re-ranking, retrieval evaluation, data governance, and enforcement of tenant and authorization boundaries throughout retrieval workflows.
• Define reference architecture and architectural principles for agentic and multi-agent AI systems , including planning, reasoning, tool use, task decomposition, delegation, collaboration, state and memory management, human-in-the-loop patterns, failure recovery, bounded autonomous execution, and communication between agents and services.
• Define architecture for identity, authentication, and authorization across AI and agentic systems , including user, workload, service, and agent identities; Role-Based Access Control (RBAC); delegated authorization; least-privilege access; scoped permissions; and secure propagation of identity and authorization context across distributed workflows.
• Establish architectural guardrails that ensure tenant isolation and authorization are preserved end-to-end across retrieval, reasoning, planning, memory, and execution , including mechanisms for validating agent-generated plans and tool interactions against applicable security, authorization, safety, privacy, and compliance policies.
• Define architecture for AI safety, security, and trust boundaries , including defense-in-depth approaches for prompt injection, unsafe tool invocation, cross-tenant information exposure, untrusted content, data provenance, policy enforcement, and secure interaction between models, agents, tools, services, and enterprise data.
• Establish architectural approaches for AI memory and state management , including short-term and long-term memory, tenant and user isolation, authorization-aware retrieval, lifecycle management, retention, deletion, provenance, privacy, compliance, and safe reuse of contextual information.
• Define architectural expectations for automated validation, testing, and architectural conformance , ensuring functional correctness, tenant isolation, authorization, security, reliability, deployment readi