Harsha Nori

Harsha Nori leads the AI team for Microsoft AI Health, where he works on language models to improve consumer healthcare. He was previously a Director at Microsoft Research, focusing on responsible AI, and has led several joint Microsoft–OpenAI efforts to measure and improve language model performance in medicine. He co-created Guidance, a widely used software toolkit for controlling language model outputs that enables structured outputs/JSON mode at OpenAI, Microsoft, and many open-source LLM inference engines. His research on interpretability and privacy-preserving machine learning has been published at ICML, NeurIPS, ICLR, KDD, CHI, and AAAI. Harsha is a graduate of the Georgia Institute of Technology.