Principal Software Engineer - AI Safety and Security
About the role
AI systems are rapidly becoming more capable, autonomous, interconnected, and embedded in products people depend on. As models gain memory, tools, identity, and the ability to act across systems, new risks can emerge through long-running interactions, coordinated agents, and behavior spanning accounts, products, and time—often beyond what any individual product or runtime control can see. Within the Artificial Generative Intelligence Security (AeGIS) team, the AI Safety Platform (AISP) is Microsoft's shared AI observability, detection, investigation, and hunting platform built for this new landscape. We connect fragmented activity across Microsoft's AI estate and turn it into reusable detections, investigation-ready evidence, threat intelligence, and customer protection. We are looking for a hands-on Principal Software Engineer for a deeply consequential technical builder role at the intersection of cybersecurity and generative AI. You will lead through code and ideas: defining detailed architectures, building large-scale data and detection systems, working through security and privacy reviews, and partnering with cross-company leaders, product engineers, applied scientists, threat hunters, and incident responders from early exploration through production operation. The ideal candidate is deeply curious about how AI systems work, how they break, and how to make them safer and more robust. We are a mission-driven, collaborative team that leans into uncertainty, learns quickly, and builds on one another's ideas. Our people bring a wide range of backgrounds, disciplines, and life experiences, and we believe that diversity makes both our engineering and our judgment stronger. We want someone energized by technically fascinating problems and by the responsibility to protect users at the scale of some of the world's most widely deployed AI platforms.
Responsibilities
- Lead the design and implementation of shared platform capabilities: guide architecture and technology decisions, establish reusable engineering patterns, write critical production code, and unblock complex technical work across teams.
- Build and evolve company-wide AI observability and monitoring systems that enable attack reconstruction, automated detection, evaluation, threat hunting, and investigation, using modern agentic development tools and AI-assisted coding workflows to prototype, evaluate, and iterate rapidly