About the Role

The Research AI Engineer II role at Microsoft Security focuses on designing and building GenAI-powered security agents to help enterprise customers understand and mitigate identity risks. Responsibilities include modernizing security workflows through intelligent automation and developing secure, scalable AI pipelines that integrate across Microsoft's security ecosystem. The position requires expertise in AI systems, large language models, and security concepts.

Key Responsibilities

• Design and build GenAI-powered Security Agents for identity risk mitigation across Entra ID
• Modernize traditional security workflows through intelligent automation and multi-agent orchestration
• Develop secure, scalable AI pipelines and MCP-based services integrating identity signals and risk data
• Prototype and deploy advanced agentic capabilities like RAG, multi-agent frameworks, and long-term agent memory
• Collaborate with Microsoft Research to drive model and signal exchange through Agent-to-Agent framework

Required Skills & Qualifications

Must Have:

• Bachelor's Degree in Statistics, Econometrics, Computer Science, Electrical/Computer Engineering or related field with 2+ years experience OR Master's with 1+ year OR Doctorate OR equivalent experience
• 2+ years hands-on experience building and deploying production-grade AI systems using Python, PyTorch, or TensorFlow
• Proficiency in large language models (LLMs), embeddings, vector databases, and retrieval-augmented generation (RAG) architectures
• Understanding of identity and security concepts including authentication, access control, detection signals, and risk-based decisioning

Nice to Have:

• Experience with Azure AI Foundry, OpenAI APIs, or Semantic Kernel for enterprise-scale AI development
• Familiarity with Microsoft Security offerings like Entra ID, Microsoft Defender, MDE, or MDI
• Knowledge of MCP (Model Context Protocol) or other interoperable agent architectures for secure model orchestration

Benefits & Perks

• Industry leading healthcare