OpenAI's New Residency Wants Physicists, Not Just PhDs

OpenAI is officially opening a new door into its labs—and it isn’t just looking for AI PhDs. The company just announced the OpenAI Residency, a six-month, salaried program launching in April 2022 that’s designed to turn exceptional researchers and engineers from other fields into full-time AI talent. It’s a direct evolution of the company’s previous Scholars and Fellows programs, but with a critical shift: Residents won’t be following a curriculum. Instead, they’ll be embedded directly into active OpenAI teams, learning by shipping real research alongside the lab’s existing staff.

The numbers behind this mentorship model are already surprisingly strong. Over the last three years, OpenAI has converted more than 20 people from similar programs into permanent technical staff—a figure that now represents one in six of the company’s entire technical team. That’s a significant pipeline, and the Residency is explicitly designed to widen it by broadening the applicant pool. They’re making a deliberate play for candidates with unconventional backgrounds, whether that’s a mathematician, a neuroscientist, or a self-taught engineer fresh out of high school.

Sam Altman, OpenAI’s CEO, framed the hunt for talent in almost agnostic terms. “This type of thinker might be at a university, they might be fresh out of high school, working at a cutting-edge tech company or building something on their own,” he said. The program splits into two concrete tracks: an AI Software Engineering path for engineers from fast-paced environments who want to tackle AI-scale infrastructure, and an AI Research track specifically targeting people with a record of achievement in non-ML fields. Ilya Sutskever, OpenAI’s Chief Scientist, put it bluntly: “There are many talented people who want to contribute to AI but cannot find an easy way to do so.”

The testimonials from former Scholars are telling. Christine McLeavey, now managing the Multimodal team, said the program gave her the confidence to jump from online courses on fast.ai to full-time research. Jonathan Ward credited his mentor and the access to OpenAI’s infrastructure for helping him “acquire a research taste” much faster. The subtext is clear—OpenAI is betting that the raw intellectual horsepower from adjacent domains, when given the right scaffolding, can accelerate its mission toward AGI more effectively than simply fighting over the same limited pool of traditional AI graduates.