Haidar Khan is a research scientist with experience at Meta, SDAIA, and Amazon, and currently serves as CTO and co-founder of ai.astrolabe, as well as a partner at KhanTeT. His work focuses on natural language processing (NLP) and efficient machine learning, particularly building smaller, faster models without excessive compute requirements.
He is a lead contributor to ZeroSumEval, a project advancing scalable evaluation of large language models (LLMs). While at SDAIA’s National Center for AI, he co-led the development of ALLaM, a top-tier bilingual LLM for Arabic and English. At Amazon Alexa AI, he worked on NLP tasks including semantic parsing and helped build Alexa Teacher Models, a precursor to Amazon’s general intelligence efforts.
Haidar completed his PhD in Data Science at Rensselaer Polytechnic Institute (RPI), where his research with Prof. Bülent Yener focused on predictive modeling in medical data, in collaboration with Mount Sinai Hospital’s Epilepsy Center.
Outside of research, he dedicates his time to family, studies Quran and Arabic poetry, learns languages (he speaks four), rides horses (western style), and enjoys hunting.
