High School Dropout Turned Self-taught OpenAI Scientist, Gabriel Quits Job to Become Founder
According to Dynasty Beating monitoring, Gabriel Petersson, the former Research Scientist of OpenAI's video generation model Sora team, and a self-taught young talent from Sweden, announced his departure from OpenAI. Petersson stated that he left OpenAI to pursue independent entrepreneurship, with plans to create his final product before the advent of Artificial General Intelligence (AGI), and made a high-profile declaration to enter Founder Mode.
Petersson's personal growth journey in the elite AI community has been truly groundbreaking. Born in a remote village in Sweden that is hard to find on a map, Petersson dropped out of high school at the age of 17, even admitting to feeling dull when he couldn't understand a machine learning course. However, he abandoned the traditional path from undergraduate to postdoc studies and quickly mastered machine learning and frontend engineering development through a top-down self-learning approach by asking ChatGPT frequent questions daily.
During his job search, Petersson, in order to avoid being screened out due to a lack of formal education, adopted a Demo-Driven strategy to bypass the recruitment process. He independently developed high-performance web image grids and table components to address the pain points of Midjourney and Dataland products, impressing the founders and subsequently joining two top teams. By the end of 2024, the 22-year-old Petersson successfully joined OpenAI as a research scientist and participated in the core development of the Sora model.
As OpenAI shut down further development of Sora in March 2026 due to computational costs and other considerations, the core creators of the team and several researchers have gradually left in the spring. Petersson transitioned earlier this year from Sora to forming a new team within OpenAI, but only a few months later, he decided to completely depart from the top-tier lab and re-enter the startup ecosystem as a founder.
