AI author, founder and independent researcher. Founder & CEO, ai021.ai, Inc.
Nearly three decades in education taught him that trust is earned by checking. Now he is building AI you can verify.
Zaf's story begins in Durban, South Africa. At the University of Durban-Westville, now the University of KwaZulu-Natal, he earned a B.A. (1989), an HDE (1990) and a B.Ed. (1992). He taught in Durban before moving to the United States, where he completed a master's in special education at Minot State University in 1993.
From 1993 to 1998 he taught at a tribal school in South Dakota. Alongside his teaching, he was a school psychological examiner in South Dakota, 1995–1998. In the same year that role ended, he completed an Ed.D. in Education Leadership & Administration at the University of South Dakota.
Assessment work leaves a lasting habit: a single test score rarely tells the whole story, and a person's judgment belongs in the picture. That habit stayed with Zaf, and it now shapes how he thinks about AI.
Next came Tennessee. He led special education for Williamson County Schools (1998–2000), directed research at the Tennessee Department of Education (2000–2001) and prepared new teachers at Belmont University (2001–2005).
From 2005 to 2022 he served as an associate professor in the Department of Elementary & Special Education at Middle Tennessee State University. In January 2022 he presented a peer-reviewed paper at the 20th Hawaii International Conference on Education on supporting human decision-making when working with AI in education. He retired from MTSU in 2022.
After retiring from MTSU in 2022, Zaf began building AI systems on his own. He brought the same questions he had asked for years about learners: how does it really work, how do you know it is right, and where does human judgment fit?
Those questions run through his books. Train Your Own LLM opens up how language models work and shows readers how to fine-tune and build one themselves. Elastic Intelligence turns to people, and to the judgment that matters as AI reshapes work. A third book, Beyond the Benchmark, is in development.
In 2026 he founded ai021.ai, Inc. and serves as its Founder & CEO. The company puts his long-held view into practice: people should be able to check what an AI tells them, rather than take it on faith.
Mission. ai021 is building AI you can verify. Our small models are designed to answer from organized, source-linked knowledge, so answers can be traced back to their sources and checked.
Vision. A world where trusting AI means checking it, because every fact it uses has an address you can look up.
His current work centers on AIOS, a vision for an AI-native operating system in which intelligence is a native part of computing rather than an app layered on top. The long-term aim is a secure, adaptive platform that coordinates intelligent tools across devices while keeping human authority and privacy at its core.
Each book takes up one part of the same idea: understand the tools, keep your judgment sharp, and check before you trust.
A friendly guide to how language models work, with hands-on chapters on fine-tuning and building one on modest hardware. It is written for readers who want to understand a model by building their own.
Fine-tuning changes behavior; retrieval supplies knowledge.
From Train Your Own LLM, Chapter 1.
Benchmarks still matter, and a fit check adds to them. The same AI can shine in one setting and stumble in another.
Zafrullah Khan, Ed.D., is an AI author, founder and independent researcher, and Founder & CEO of ai021.ai, Inc., which is building AI designed so answers can be traced back and checked. After nearly three decades in education, he wrote Train Your Own LLM and Elastic Intelligence (2026).
Zafrullah Khan, Ed.D., is an AI author, founder and independent researcher, and the Founder & CEO of ai021.ai, Inc. ai021 is building AI you can verify. Its small models are designed to answer from organized, source-linked knowledge, so answers can be traced back to their sources and checked. His current work centers on AIOS, a vision for an AI-native operating system that keeps human authority and privacy at its core.
He is the author of Train Your Own LLM, a hands-on guide to language models, and Elastic Intelligence, on human judgment alongside AI, both published in 2026. Before turning to AI, he spent nearly three decades in education. He taught in Durban, South Africa, before moving to the United States, where he was a school psychological examiner in South Dakota, 1995–1998, and an associate professor at Middle Tennessee State University, 2005–2022. He holds an Ed.D. from the University of South Dakota.
ai021 is building toward a world where trusting AI means checking it. The mission, vision and AIOS vision above describe where the company is headed.
Whether it's an interview, a speaking invitation, a partnership idea or a question about the books, Zaf would be glad to hear from you.
Statements about future plans are aspirations, not promises or an offer of securities; actual results may differ.
