Interviewing at OpenAI as a new grad is genuinely different from most FAANG-style processes, and I didn't find a lot of specific advice when I was going through it. Here's what I learned, some of it the hard way.
They do hire new grads, but it's targeted. Mostly through the university recruiting pipeline or if you've done a research internship there. If you're applying cold from the career portal as a new grad, conversion rates seem low. Having an intern return offer or a referral matters more here than at Google where volume is high.
The coding bar is real. For entry-level SWE roles I heard consistent reports of 2-3 coding rounds at medium-to-hard LeetCode difficulty. I saw two mediums and one hard in my onsite. Know your trees, graphs, DP at minimum. They're not trying to trick you but they are filtering for strong fundamentals.
Don't skip the behavioral prep. This surprised me. Even for an entry-level role they run a full behavioral round and the values alignment component isn't soft. They asked me things like: how do I think about the societal impact of technology I build? what does responsible engineering look like? I had a 45-minute conversation about this. Come with real thoughts, not rehearsed answers.
Study system design even as a new grad. Not at the senior level, but they'll ask a lighter design question. Know how to think about scale, APIs, basic database choices. Walk through your reasoning out loud.
The timeline can be slow. Mine was 8 weeks from application to offer. Don't assume silence is a no.
For comp as a new grad: I've seen reports around $180-200k total depending on the team and offer negotiation. Equity matters more here than almost anywhere given the stage. Get competing offers if you can and use them.
Happy to answer anything. Good luck.