OpenAI puts $7.5k/month behind diversity in deep learning

OpenAI just threw open the doors to its first Scholars program, and the math is refreshingly direct: six to ten people get $7,500 a month, plus mentorship from some of the sharpest minds in the field, to drop everything and study deep learning for three months. The application window opened immediately and closes March 31st, with decisions landing by April 16th. It’s a remote gig, so long as you’ve got US work authorization and are in a US timezone — though a desk in the San Francisco office is available if you’re local.

The premise is simple enough. OpenAI wants individuals from underrepresented groups who can commit full-time from June 4th through August 31st. In return, they’ll get a mentor checking in weekly via video call, a group Slack to trade questions and frustrations, and $25,000 in AWS credits to actually build something. The one hard requirement: you need to open-source a project by the end. That’s the price of admission — real work, publicly documented.

It’s worth reading between the lines on why this exists. OpenAI’s own fellowship and the usual residency programs demand existing machine learning chops and, often, a move to the Bay Area. That’s a non-starter for plenty of people with families or other obligations. The Scholars program explicitly lowers that bar. They aren’t asking for a PhD — the application itself notes that many OpenAI employees don’t have one. What they want is someone who can follow a calculus and linear algebra refresher and is comfortable writing Python. The rest is about drive: can you blog weekly about your progress, and will this grant unlock something you couldn’t reach on your own?

Selection hinges on a mix of personal impact, self-motivation, and technical baseline. The stronger your foundation, the faster you’ll hit the deep learning material itself. Mentors are drawn from both OpenAI and the broader community, with AWS underwriting the credits for external mentors. It’s a targeted bet that funding focused, documented effort from a wider slice of people will do more for the field — and for AI’s ability to serve everyone — than another batch of identical resumes ever could.