Aashish Kolluri

À propos

I am an AI security researcher at Microsoft Research in Cambridge, focusing on building most advanced secure and privacy-preserving AI agent systems for safe deployment. In the Confidential Computing group, I contribute to Project Roma, which adds deterministic protections to AI agents to keep them safe and confidential, even when models act unexpectedly. My works also address defending against data poisoning and backdoor attacks, developing privacy-aware and provably secure machine learning systems. I combine my experience in security, distributed systems, and AI to create the technical foundations for safe, reliable autonomous AI.