Systems, cloud, AI, and writing
CTO, Deputy CISO and Technical Fellow, Microsoft Azure.
Mark Russinovich’s current work spans AI research, the podcast, technical writing, and hands-on projects — with books, long-form material, and his drawings collected here.
Security and practical AI
Current work centers on agent security, model robustness, and dependable cloud-scale systems.
Latest paperMetaBackdoor: Exploiting Positional Encoding as a Backdoor Attack Surface in LLMs
Latest episodeScott & Mark Learn To… Double-Check References
Scott & Mark Learn To
The latest episode starts with thousands of hallucinated citations Mark found in conference submissions, then opens out into accountability, taste, and where leaning on AI quietly costs you judgment.
Latest episode: Double-Check References (Ep. 39)
AI and agent security
Recent papers focus on agent security, model robustness, unlearning, and dependable AI infrastructure, with direct links where public versions are available.
Books
Mark’s books span the Jeff Aiken cyberthrillers, Windows Internals, and Troubleshooting with the Windows Sysinternals Tools.
Blog
Long-form essays, interviews, conference sessions, and technical commentary from across the archive.
About Mark
Biography, career background, books, and the path from Windows internals to cloud architecture and AI research.