Gabriel Peterson grew up in a remote town in Sweden. He never finished high school. One day, his cousin called and said he'd met a brilliant researcher in Singapore who had an idea for an AI-powered product recommendation system. Gabriel got on the next bus to Stockholm and never went back to school.
That was the start of an unlikely journey. With no formal education in computer science or mathematics, he learned by doing — building recommendation systems for e-commerce companies, cold-emailing prospects, showing up in person with printed A/B test results. He learned on the job, the hard way.
Then came ChatGPT. For Gabriel, it wasn't just a tool — it was a personal professor. He used a 'top-down' learning method: start with the problem you need to solve, then fill in the knowledge gaps as you go. When he hit a wall with math, he'd ask ChatGPT to explain it. When he needed to understand a new machine learning concept, he'd work through it conversationally rather than through a textbook.
His Stack Overflow answers, accumulated over years of troubleshooting, ended up helping him qualify for an O-1 extraordinary ability visa to the United States — 'academic publications' counted toward the criteria.
Today, Gabriel Peterson is a research scientist at OpenAI working on the Sora video generation team. He spends his days observing model outputs, finding artifacts, forming hypotheses about what to change in the architecture or training data, and running experiments.
He describes the research loop as 'surprisingly simple and pure': look at the results, find what doesn't look right, propose a hypothesis, retrain the model, and evaluate by staring at the output. When he needs fresh ideas, he feeds the entire codebase to ChatGPT and asks for ten improvement suggestions or the latest relevant papers.
His advice to others who feel behind or underqualified: high initiative is your passport. Don't wait for permission or credentials. Find a real problem, start solving it, and use AI as your tutor and partner. College no longer has a monopoly on knowledge — the definition of 'expert' is being rewritten.