- Role
- Co-founder of NEAR Protocol; CEO of the NEAR Foundation since November 2023; leads NEAR AI
- From
- Kharkiv, Ukraine
- Studied
- Applied mathematics and computer science, Kharkiv Polytechnic Institute
- Before NEAR
- Salford Systems (from 2008); Google Research (2014 to 2017)
- Known for
- Co-author of "Attention Is All You Need" (2017)
Illia Polosukhin
Polosukhin grew up in Kharkiv and has said his interest in artificial intelligence started in childhood. He competed in programming contests at school and university, took a master's degree in applied mathematics and computer science, and moved to California. He spent about six years at Salford Systems, a predictive-analytics software company in San Diego, building data-mining tools.
In early 2014 he joined Google Research, where he became an engineering manager and worked on the TensorFlow machine-learning framework and on question answering for Google Search. That placed him in the group that produced the Transformer paper in 2017. He is the last of the eight listed authors, credited in the paper with early implementation work.
He left Google the same year to start NEAR.ai with Alexander Skidanov. When program synthesis proved too early, he led the pivot to building a blockchain, co-designed Nightshade, and has been the project's most visible figure ever since. After Russia's full-scale invasion of Ukraine in 2022 he helped set up Unchain Fund, a crypto-based relief effort.
In November 2023 he became chief executive of the NEAR Foundation, and within two months oversaw a 40% staff cut that he framed as narrowing its focus. From 2024 he steered NEAR back toward AI under the banner of "user-owned AI": the argument that powerful AI should be open, verifiable and controlled by its users, not by a few corporations. His recurring public prediction is that AI agents will become the main users of blockchains, with AI as the interface and blockchains as the back end.
Deep dive: how to weigh the "Transformer co-author" credential
It is real and significant. Very few people in crypto have comparable standing in AI research, and it gives NEAR access to audiences, such as Nvidia's GTC stage, that most blockchains cannot reach.
It also has limits. Polosukhin left frontline AI research in 2017. NEAR AI's products mainly host and protect models built by others; critics note that it has not published frontier models of its own. And a founder's reputation says little about whether a token will hold its value. A sensible reading is: strong evidence of technical ability, weak evidence about commercial outcomes.
Deep dive: one person, many hats
Polosukhin is at once co-founder, Foundation CEO and the leader of NEAR AI. That concentration gives NEAR clear direction, which many decentralised projects lack. It is also a "key person" risk, and it sits uneasily with claims of decentralisation, a point raised during the 2025 inflation dispute. The Foundation has said its role should shrink over time.