Principal Software Engineer, Perception Platform and AI Systems

** 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 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 Software Engineer
** to provide technical leadership for the evolution of the Perception Platform and drive foundational AI and platform investments across Microsoft Security. You will define architecture and engineering strategy for sophisticated Generative AI, Retrieval-Augmented Generation (RAG), memory, reasoning, agentic and multi-agent systems, identity and authorization, AI safety, and distributed cloud platforms operating in security-sensitive enterprise environments.

Combining deep distributed systems and cloud platform expertise with strong AI systems architecture experience, you will establish architectural patterns that enable multiple engineering teams and Microsoft Security products to build secure, reliable, compliant, and scalable AI-powered experiences on common platform foundations. You will operate across organizational and product boundaries to translate ambiguous business and product requirements into durable technical architectures, drive alignment among senior technical leaders, and identify opportunities for MSEC-wide engineering leverage.

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 architecture and long-term technical strategy
** for enterprise AI platform capabilities spanning generative AI, retrieval, memory, reasoning, agent orchestration, model interaction, execution, evaluation, extensibility, security, and safety.

+  
** Architect production-grade AI and Retrieval-Augmented Generation (RAG) systems,
** including data ingestion, indexing, embeddings, semantic and hybrid retrieval, context construction, grounding, provenance, re-ranking, and retrieval evaluation.

+  
** Design enterprise AI memory and knowledge architectures
** that manage conversational context, durable memory, organizational knowledge, and retrieved information while addressing authorization, provenance, freshness, retention, correction, deletion, and compliance requirements.

+  
** Architect agentic and multi-agent AI systems
** that support planning, reasoning, tool use, task decomposition, delegation, collaboration, state and memory management, human-in-the-loop patterns, failure recovery, and bounded autonomous execution.

+  
** Define identity, authentication, and authorization architectures for AI agents,
** including user, workload, service, and agent identities;
Role-Based Access Control (RBAC); delegated authorization; least-privilege access; multi-tenant isolation; and secure propagation of identity and authorization context across distributed workflows.

+  
** Establish security, safety, and execution guardrails
** that protect AI systems against prompt injection, malicious or untrusted content, unsafe tool invocation, data exfiltration, privilege escalation, unauthorized delegation, excessive agency, and cross-tenant access.

+  
** Design AI architectures that enforce authorization and trust throughout retrieval and execution,
** ensuring agents access only permitted data and that AI-generated plans, decisions, and tool calls are independently validated against applicable security, authorization, safety, and compliance policies.

+  
** Establish end-to-end reliability, evaluation, and observability frameworks
** for distributed AI systems, covering retrieval and model quality, grounding, agent behavior, authorization decisions, tool execution, task completion, safety, telemetry, regression detection, failure recovery, and graceful degradation.

+  
** Partner across engineering, research, product, security, privacy, and compliance organizations
** to translate…