Careers at the Frontier: Hiring the Future at OpenAI

Welcome to the OpenAI Forum. I'm Natalie Cone, our community architect. Today's conversation is for the curious builders, the problem solvers, and those wondering what it really takes to build and join the most mission-driven AI lab in the world.

Our mission at OpenAI is to build AGI that benefits everyone, and we typically feature conversations in the OpenAI Forum that highlight how our technology is helping people solve hard problems. But today, we're switching gears a bit and taking a peek behind the curtain at OpenAI to understand more about the heart of the organization, the recruiting team responsible for hiring the people that make our company, and the person recently tapped to helm that team.

I'm thrilled to be joined by Joaquin Quiñonero Candela, our head of recruiting at OpenAI. In addition to getting to know Joaquin a little bit better tonight, I'm also excited to share that we've engaged Joaquin's team in preparation for this fireside chat.

Some of the advice and tips we'll share were sourced directly from the experts on the ground. We hope to address some of the burning questions you all have in relation to working at OpenAI and perhaps even navigating a career transition and new beginnings in any domain in this era of rapidly advancing technology.

It's inevitable that we won't get to everyone's questions, but Joaquin and his team have generously offered to return to the forum over the next few weeks to ensure we address as many of your questions as possible and more on that in the future.

So without further delay, I want to welcome Joaquin to the stage.

Hi, Joaquin. Hey. I was just looking for you on mute. Good to see you. Yeah. Great to meet you, everyone. Pretty thrilling to have. How many people do we have, Natalie? 11,200.

So, the largest event the forum has ever hosted, you basically broke the internet, Joaquin. And the funny thing is, when I invited you to do a talk in the forum, you were an intern, but more about that.

So, we didn't know that you were going to take the forum. So, thank you so much. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you. Thank you.

So, we didn't know that you were going to take the helm as head of recruiting when we invited you, but it did end up in a really fascinating twist of events, being a real resource for the community and the guests here. So, thank you so much for coming tonight.

My pleasure. Joaquin, you are someone who shows up with both technical depth and deep care for people and the culture we're building at OpenHouse.

AI: And I wanted to open with how we met, actually, which was back in February. And I always knew about you because the OpenAI Forum experts, they worked with your team when you were leading the preparedness team to create post-training evals, but I didn't know you personally until I was about to host Music is Math, an in-person event at the OpenAI Lab, where an AI startup, an AI founder, but also a classical composer, took NASA's 2024 solar eclipse data and rendered it into a beautiful classical composition. And we had the SF Symphony there. We had our own Christine McLeavy on piano. And you registered for the event out of the blue, which was so sweet because you're busy and the preparedness team is always swamped. All of that work is on the

critical path to releasing our products. But without even a special invitation, you registered and you showed up and you helped us get Tony Javara, your good friend who at the time was working at Spotify. You helped us convince him to wear an emotive headset so we could watch his brain on music during the event. And that really stuck with me because you showed up for a program that had nothing to do with your team. And you made me feel seen as a colleague that's not a member of the leadership team. And you helped me host our guests.

So when I saw your LinkedIn post about becoming an intern, I thought, wow, Joaquin is fascinating. We have to host him on the forum and here we are.

So maybe we just jump right in because we don't have a ton of time and we have a huge talk track.

to get through. So Joaquin, you've had a fascinating career. Your background is in applied machine learning. You previously served as technical leader at OpenAI on the preparedness team. Then you took a role as an intern, and you did that for several months. But now you're head of recruiting. So please tell us a little bit about your career and how you came to lead OpenAI's recruiting team.

How far do you want to go, Natalie? I can talk about my undergrad. I studied telecommunications engineering in Spain, nothing really to do with AI in principle. Let's start there, because I think, as you know, the forum is composed of academics from all disciplines and people looking to make career changes and break into AI. So let's just start at the beginning.

Awesome. Yeah. No, I studied telecommunications engineering in Spain, and what that was is it was a lot of electrical engineering, designing antennas, communication systems, but also electronics discrete and integrated.

And the path to AI was in telecommunications engineering, you need to do a bunch of signal processing. You need to, for example, understand if you're holding a mobile phone inside a building, you need to model how that signal is bouncing on the walls and so on, and you need to build a model of that to be able to undo that noise, that interference.

And the professor who was teaching out these things was one of the pioneers of neural networks in Spain, as it happens. And so he introduced me to that, and then I never looked back. I was hooked. Like the fact that you can build adaptive systems that can learn from data was just mind blowing to me. Beautiful. And so...

But and then you went on to get a PhD. Yeah, yeah. I got a PhD. I started it in Denmark at the Technical University of Denmark and finished it at the Max Planck Institute in Germany where then I did a postdoc.

All of that focused on machine learning, on what we called Bayesian inference and nonparametric methods, things like Gaussian processes for people who have worked on those things. Yeah.

And then after that, I went on to be a researcher at Microsoft Research in the old Cambridge in the UK where that was like my transition into industry actually because Microsoft Research was like an interesting bridge where you could still be a researcher but you were exposed to products as well. And I started to work with a Bing team when Bing launched in 2009 on ads relevance, on using machine

learning to make ads less irrelevant or more relevant, if you will. And then, yeah, that was like the big transition to then, you know, joining Facebook where it was for, for 10 years, building out the applied ML team for the entire company and then LinkedIn, where I spent two wonderful years as a, as a technical fellow working on things like AI fairness. And so AI fairness has, has kind of been a common thread for you as well. Safe AI, AI fairness, and then you worked as the head of preparedness at OpenAI.

Yeah. That one was not on my bingo card. I have to say my journey at OpenAI has been probably one of the craziest in my career in many ways. You know, I've been here for less than a year and a half, I believe. I'm not going to try to do the math. And I've just had like, like a lot of different roles. It's just like every quarter here is like a year or somewhere else, but yeah, I first

Yeah, sorry. I first joined OpenAI as a, I joined a safety systems team to work on human-AI interaction, a very focused effort, a tiny team.

And then the opportunity or the ask to come, hey, help us with preparedness. Preparedness, for those of you who wonder, is an amazing team that focuses on understanding potential catastrophic risk from AI, meaning can someone go and create a new type of bioweapon that we've never seen using AI or can someone break into a system using a sophisticated cyber attack powered by AI, or even worse, can models deviate and start having their own objectives that we can't control?

But what I really loved about that team is it really took a very quantitative and scientific approach at that question.

of vibe-based discussions on those things. And the question is, can you actually design tests to provide more like a quantitative assessment, which is what we used to do. So yeah, I did that for a bunch of time. It was a wonderful experience. And yeah, and then I became an intern.

And we have many new members of the forum and even more new guests. But for those of you who are new, I just want to remind folks that this community, the OpenAI Forum, it was launched two and a half years ago with only 200 people. And they were experts operating at the highest expression of their field. And the projects that we worked on, like so many of them were actually in collaboration with your team, Joaquin, to create those very niche data sets that would evaluate our models to see if they could potentially do harm in the world. And so the roots of this community.

are actually in deep collaboration with that team. So that's awesome, and I just want to make sure that the new folks here knew that. Thank you. Thank you, community.

But then in a very strange turn of events, Joaquin decides he wants to be an intern and he wants to go back to coding. So tell us a little bit about what inspired that transition and what that experience was like at OpenAI.

It's something that started brewing probably 15 years ago, if I'm honest. I transitioned to management in the summer of 2009. I still remember because it was a traumatic experience. All of a sudden, I'm responsible for a group of people, I don't know what I'm doing, I need to set roadmaps, coach people, hire, and all these things. I feel like as I've been on a journey of

of leadership and management, my imposter syndromes kept increasing over time because I felt I don't really know what I'm talking about. I go to all these meetings and I say words and my lips move and I'm using jargon, but I don't really know what I'm saying because I haven't built these things. I haven't worked on them. And then I felt like I was just drifting over time. And then I think end of last year, as I was wrapping up my stint on preparedness, I sort of thought, maybe it's time. Maybe it's time for me to overcome my fear. And maybe it's time to ask for what I really want.

So I had the most hilarious conversation with Sam. Sam is like, hey, we need more leadership. Are you interested in this other big?

bigger role. And I'm like, actually, can we time out here for a second? I have something crazy to say. Sam likes crazy things. And so I said, you know what I really want to do is I want to be an intern. He's like, what do you mean an intern? And I'm like, well, you know, like an intern, where you just work on stuff all day long, right?

So yeah, I went and I did an amazing three-month internship with our health team, working on evals to understand how good our models are at responding to real-world health care-related queries. I also did a refactor, simplified a little bit of our codebase. And I really got in the flow. It was an exhilarating experience. It was wonderful.

But now you're the head of recruiting. And so how did that transition happen?

I mean, I definitely, from the way that you show up for us, can absolutely see it. But for you and for the company, how did that transition happen? And what have your first few weeks in the role been like?

I can tell you, I didn't see it at all, right? I never thought I'd be in this role. Our chief people officer, Julia, and I had dinner and I was super excited to tell her about all the work I was doing as an intern on the code that I was writing, you know? And she looks at me, she says, yeah, we need a head of recruiting. And I'm like, oh, interesting. Yeah, I have a bunch of really good referrals. She's like, no, no, we need a head of recruiting. And she was looking at me in the eyes and I'm like, sure. And so here we are. And I think the thing that, you know, it took me maybe, maybe it took like a week to really think about this a little bit, because.

it just caught me by surprise. And I think what helped me make the decision was a deep sense of mission in many ways, right? To understand that although OpenAI is about 10 years old, give or take, in many ways, we're only getting started now, right? And I feel like we're at the beginning of building a company that is probably gonna be generational.

And so I thought, well, what do I wanna do, right? If this is a stepping stone for me to go and take another technical leadership role somewhere else, then maybe it's a crazy move, right? Like why would I become head of recruiting? But if I see myself here for a while and I feel like, oh, I wanna look back five years from now

and say, hey, I did my little contribution to help build this place, right? In a way where like the culture that we build and the people that we bring are really aligned with our mission and what we're here to do. And then I thought, then it became a no brainer. Then I'm like, I'm in, let's do it. Love that.

And it's the perfect segue to this next burning question for me, which was, is there a common thread that draws people to open AI? And what have you heard most often when people share why they wanna work here?

Yeah, I think there's a lot of reasons why people join open AI, of course, but I would say they probably group into four big categories, right? The first one is the mission. It's very inspiring to talk to candidates who look from the outside.

And unlike me or others who get a bit lost in the fact that we're too close to the technology, right? Like we're like, you know, sure we have models now that can do deep research and whatnot. It's like business as usual.

But if you're looking at it from the outside, I think people recognize the magnitude of the historical moment in many ways, right? We're building AGI and we're bringing it to humanity in a way that's responsible and beneficial. So I think mission is a big one.

The second one is the culture. We are a culture. We often say we're a culture of doers who do things, right? We're doers who do things. With autonomy, with agency, there's no sitting around asking for permission.

And one thing that I always find inspiring is often I'll talk about a problem that is super important.

I don't have time to work on it. And then I'll bump into someone on the corridor a few days later, and then they'll say, oh yeah, by the way, this thing you talked about, I've been working on it, right? And one example is this forum actually, where like, you know, our comms team, you know, Kayla and others, I had sort of expressed my stress, you know, and then a couple of days later, I bump into her in the corridor and she's like, oh, Joaquin, by the way, we've been preparing, we've been working with Natalie on a talk track, like we've got you. So you have like countless examples like that.

So the dual culture, I think, is amazing. The third one would be just the cutting edge, right? Like people that are attracted to the cutting edge. This is a place where we're doing insanely innovative research, but this is research with impact. And it's very rare to find such a research environment that is coalesced, galvanized, that is really,

focused on impact with obviously absolutely world-class compute, right? Let's be real. This is very important as well.

And then I would say the last thing, but not least, probably one of the most important things is people. It's the people you work with. You know, we often say good people draw good people.

And people want to work with like-minded people. People want to work with people who are a little bit like frontier people. People who, you know, want to jump at the deep end and into the unknown, who are afraid, right? Who are dealing with fear.

But if you're not the only one dealing with fear, it's kind of easier, right? Definitely.

And I actually, I've shared this with you before, but I'll share it with our friends here today. When my family or past colleagues that are trying to break into...

AI, ask me what it's like to work here, I always describe it as what I would imagine it's like the Olympics of professional work. I feel sometimes so very pushed to my limits of what I'm capable of, but you're just surrounded by brilliant people and really positive attitudes and that mission, and you actually also reminded me of my first day at OpenAI, I'm so glad that my experience like yours was anchored first in the research org, because you learn so much about OpenAI in the research org first, and my skip level boss was Wojciech, one of our co-founders, and he took me out for coffee, and he's a very fast walker, and we're walking really fast in the mission, and he asked me why I wanted to join OpenAI,

And my background is different. I'm not a technologist. So I was on the outside looking in. I had a little bit of experience at scale AI, in data operations, but also in building machine learning community. And I felt like having a different, a very different background and vantage point would be so useful in building a community of collaborators to help us build the technology. Because I wasn't gonna take anything for granted.

So I agree with all of that. Everybody is mission driven. The team is like a world-class Olympic team. And we're all afraid. And I think earlier we spoke, it's not about being fearless. It's about, we're the kind of people that are very afraid, but we do it anyways. And then we jump into support.

support each other. So all of that resonates. Yeah, yeah. And I think there's a story that I love from my internship. It wasn't a piece of cake. I was terrified. I mean, picture this. I had not touched code for 15 years. And even when I touched code in the past, it was like researcher code. I grew up using MATLAB, you know? And so I was not a good programmer ever in my life. And, but I was determined. I really wanted to do this. And it's really the kindness, the immense kindness of the team. My mentor, my manager, the head of our health team, were incredibly patient, dedicated.

And then the way they would celebrate, you know, my first pull requests. My first pull requests were very simple. I was checking in documentation. But then later on, I went to like, do a pretty massive refinement.

many thousands of lines of code worth of a PR. And so that combination of like non-judgment and almost like support for the man in the arena in some ways, right? Like the fact that you show up, people really love that. So you want to be in an environment like that. It was incredible. I love that you got that experience, but then we were able to drag you back into leadership again, because you are exactly where you belong, because- I'll try to get out, I'll try to get out. You are what we call a culture carrier at OpenAI. And I think the very first example of showing up for the Music is Math forum event. Also, after the Music is Math forum event, you stayed in touch with me and you made sure that I was meeting other people, other women leaders, because I told you that mentorship was very important to me. I like to anchor to other experienced women.

who have also faced adversity, to help guide my path. I really believe in mentorship. And you connected me with Irina Kaufman.

So you are a culture carrier, and you set the tone of our organizational culture, but why is being aligned to our mission and culture so important?

It's just because of what we're building, right? Because of the responsibility we have. We're building technology that will change the world, and there's no blueprint for how to build it. And there's no blueprint for how to deploy it responsibly and in a way that's beneficial.

So you need crazy adaptability and agency. And one quote I love, which doesn't really seem to have a good attribution, people argue, is culture, it's strategy for breakfast.

And culture is what people do when you're not looking. So if you really want to build a team, a large team that scales, filled with people who are builders and doers with agency and autonomy, the only way you get all this together is through culture, a culture of fearlessness and responsibility.

And if you really want to build beneficial AI, the focus shouldn't be on the AI, the focus should be on the humans. I want to share a story. I went to the NeurIPS conference, which I attended for 20 years in a row or something like that. And I think this was 2018 or 19. There was a workshop on responsible AI that invited Yo-Yo Ma, the cello player. And Yo-Yo Ma,

was on stage and he was being interviewed and someone asked, hey, Yoyo, as AI gets more powerful, what would it take for you to trust the AI? And Yoyo took a deep breath, he's a very thoughtful person, and said, well, I would need to know who built it. I would need to know who are the humans behind that AI, what are their intentions, and what are their values. And that quote has really, really stuck with me. That makes a lot of sense.

I actually, on another Yoyo Ma note, worked really hard to try and get Yoyo Ma to be on the stage at Music is Math, but that fell through. But thank you so much for sharing that story with us, Joaquin, it's really lovely. I could not agree with you more. And I do feel like every single one of our colleagues, though we might not agree on every.

day-to-day move that we make, we all agree that we're building this technology, not just for the sake of building the technology, but because we're so excited to see how it can help people and solve our hardest problems.

So maybe we'll shift gears just a little bit, and maybe you can talk about what a typical day in the life of your recruiting team looks like.

Oh, a typical day is crazy. It's like crazy all the time, right? We have a constantly changing environment. The goals are always changing. We can't really predict what products we'll be shipping long-term because the models keep getting more capable, right? So it's very hard then to know how much do different teams need to grow, what's the headcount planning, how do we staff the team correctly, and all these things.

At the same time, we're very blessed and incredibly grateful.

for having a huge amount of interest. Every time we post a new job description, there is a massive amount of applications where this causes certain challenges, of course, it's hard to keep up, but at the same time, we're incredibly grateful for the amount of interest.

The other thing that I wanna say is the team is amazing. The recruiting team is phenomenal. You have such a diversity of backgrounds. People come from technical backgrounds, from the arts, from all kinds, and have all kinds of crazy hobbies.

And one thing I want to tell you is, is it was super exciting to get to know the team. When I joined, I really wanted to do a listening tour. And naively, I thought, you know, I'll meet with everybody. But of course, that's very difficult to do in a short amount of time.

So one of our managers on the team,

team, Chrissy Robinson, if you're there. Hey, Chrissy, thank you. She came to me and she's like, well, why don't you use ChatGPT? I'm like, well, what do you mean? She's like, have ChatGPT interview everyone on the team.

And I'm like, well, I'm the one who wants to know how they feel. She's like, no, no worries. We'll create a template. We'll create a prompt that has a questionnaire. You can give me the questions. And then we'll ask people to use voice mode with that prompt and have ChatGPT interview them with those questions. And then have ChatGPT produce a transcript of it.

And then we'll have people paste it in a Google form and send you all of the transcripts. You can already hear product opportunities here, left and right. We could do much better than this, but hey, that's how we did it. So we got back 75,000, almost 75,000 words of feedback from the team. The response rate was

very high. And initially, I thought, well, let me use chat GPT to summarize a lot of this. But the problem is, it was kind of like glossing over important details. And I can't really blame it, right?

Because often if you if someone sends you, I don't know, like 1000 words of feedback or something like that. It's often like in the tone, it's in the small things, right? In the nuances.

So I basically set out to read it all. And I took, I don't know, I took like 10 or 12 hours to read every single word. And and do my own sort of summary and themes.

And then we had a team on hands. And I kind of like read back to the team, everything that I had heard. And that, for me, was a magical moment of connection with with the team.

And it really helped me appreciate how different everyone is. But at the same time, how connected around two key things, but like everyone on the team is incredibly connected to the mission.

and everyone on the team is incredibly scrappy, right? They'll say, look, we're flying the airplane as we're building it, the engine's on fire. Someone's asking, why are the snacks not better? You know, here we are.

And then the teamwork, yeah, and then the teamwork, right? So I was also pretty nervous about this event today. I don't think I've ever spoken in front of 11,000 people. I think my record was maybe four or 5,000 that F8 when I was at Facebook on stage, but 11,000, that's a new thing for me.

And some of you might be influencers and you have like hundreds of thousands of things. That's not me. I'm like, this is a world record to me. And the team jumped into action. So the team are like, oh, let's actually help you prepare. People are gonna wanna get advice for, you know, how to interview here or like what kind of people are you looking for?

or all these things, we're going to help you. And I'm like, well, you better help me, because I'm in week five of being a recruiter, okay? So all of you are better than I. So I'm going to, I'm actually going to embarrass people. I'm going to read the name of everybody who got together to help, to help prepare for this. It was like, they prepare like a 10 pager, you know, and we're going to try to kind of clean it up and maybe make it available with the community somehow.

But yeah, like Yarin Strasma, Madeline Altman, Brian Hurd, Nick Martin, Jordan Williams, Sean Baker, I apologize, Corey Collins, Sarah Warko, Jordan Smith, Jared Long, John Lamb, Brad Bigler, and Devin Lloyd. And last but not least, Red Avila organized the whole thing and sort of jumped on it. And while I'm reading names, we need to also talk about all the people who are behind the scenes, right? Natalie Cone, thank you for doing this. Selena Ma, and I'm sorry, I'm reading down in my window because I made some notes on my other screen.

Charlie, Etienne, Morgan, Crossland, Kayla Wood, Matisse, Azweta, Scott Etheres-Smith, Natalie Cone, I'll have you twice, and one for all the work and one for being here, and then Caitlin Maltbie, and of course, my incredible executive partner, Crystal Boyd. I might have forgotten someone, I'm very sorry, but these names, the people, this is everything. It is everything, and we're so lucky to have you.

You're giving me goosebumps right now, Joaquin, and I did have the pleasure of working with your entire team to curate this talk with you, and it was fascinating.

I mean, Devin provided so much excellent feedback, the synthesizing, all of those pages. Like, I've also never been a recruiter, and you invited.

me as an honorary guest because I was learning about recruiting in order to host this event with you. And I just, I mean, I can't believe I was so blind to this huge team at OpenAI, which in my intro called it now, I believe it's the heart of OpenAI, all the work that you guys do and all the insights that your team has was fascinating. And yeah, I agree. Last but not least, Red Avila jumped in and organized all of the information in a way that you and I weren't even thinking about. So thank you so much to that team. And I, I truly hope that this is just the beginning of our collaborations together because it was pure joy to put this together with you guys. All right, let's keep moving.

So Joaquin, this is like perfect for you. I think this is what the community is going to want to know. But given your

technical background, how do you see recruiting teams now using AI to make the processes and the work more efficient and more effective?

Yeah, the most important thing here is recruiting is human. It's humans recruiting humans. Recruiting is people-centric. AI is not going to replace recruiters. Humans are going to be making the decisions. But AI can actually both make the work a lot more efficient and actually it can work very strongly in favor of candidates too. I want to give a couple of examples.

Yes. So if there's any sourcers or recruiters out there in the audience, maybe you know tools like Juicebox that helps surface non-traditional candidates using AI. In fact, you can use tools like

like OpenAI's deep research as well, because not everybody has like a well-curated LinkedIn profile, but people can actually be active on different forums, right, on X, on Reddit, on GitHub. People can have, you know, written papers and notes and things. So there's ways in which AI can actually make the process of going and finding and almost like completing profiles for candidates who don't have like a well-curated, you know, LinkedIn profile or well-put-together CV.

In fact, when I was at LinkedIn, the team worked on something really interesting, which is they'd kind of look at someone's profile and they try to sort of look at their activity and stuff and try to almost like suggest, hey, you can make your profile more complete. And if you, you know, we think you might actually have these skills

haven't listed, either list them. But if you don't have them, if you were to acquire them, we'd be able to match you with a lot more jobs than right now. And that was actually powered by AI as well.

And then you have the mundane stuff that is actually very important as well. I'm religious about interview feedback. Interview feedback has to be complete, high quality. It has to be in within a couple of hours of the interview. We need to move fast.

But it's tedious and it's work. So there's tools out there that both transcribe an interview and or help summarize the notes and really assist the interviewer with interview feedback. And actually, I'm pretty hopeful about these tools. I think they can help make the process a lot fairer as well and ensure more apples to apples in many ways.

And there's a lot of more work to come on using AI to improve.

candidate experience, which is something that's close to my heart and where right now, in all honesty, we're we're struggling to get back to everybody. And I'm really hoping that we can lean into AI to help us do that better.

That's a beautiful goal. And one of the impetus to bring you here today and to collaborate with your team was so hopefully we can acknowledge that to everyone that's listening, that we are so honored to have all the interest and we're really doing our best, but we're we're struggling, as you said, we're struggling, but we are we're powerful and we're definitely going to get there. And I think it's very exciting that we have a technical leader now, the head of recruiting.

OK, we have a few more minutes. We actually have we have 13 more minutes. And I want us to.

dig into what I think is probably incredibly top of mind for everyone here, and that is, what does it actually take to work at OpenAI?

So Joaquin, is there a common denominator among the people that we hire and what do we truly look for in candidates?

Yeah, I'm gonna play back a summary of what the team, all of the names that I mentioned earlier summarize and put together, complemented with some of my own experience, of course.

I think the overarching theme, I can't speak anymore, the overarching theme that brings everything together is that we really look for doers who do things, right? And maybe if I have to, I don't know, come up with,

with a bunch of key traits. I'd say we look for owners. We look for people that have demonstrated an ability to solve end-to-end problems with agency and autonomy, and who are comfortable in messy and changing situations, because our environment is changing all the time. So ownership, end-to-end ownership, and agency, and autonomy.

The second one, I said it, but I cannot emphasize it enough. Being a builder, having a scrappy zero-to-one mentality. We often say, at OpenAI, everyone is an operator.

The third one would be curiosity. We look for explorers. We look for people who challenge the status quo, people who see beyond the horizon, people who are always learning.

Next one, I would say, and these are not in order, would be motivation. Like, why do you want to come here, right? And they're often, look, it's less about are you an AI expert, right?

You hear people in the community probably know about things like our residency program, where we've even brought in researchers to work on AI, even though their background was in physics or in biochemistry or in other things, right?

And one of them, Joaquin, was even a professional poker player. True. We have all sorts. We have all sorts, without a doubt.

But yeah, that motivation, right, that willingness, and obviously this is combined with a curiosity, right? That curiosity and motivation, I think, are key. We look for excellence. We look for a repeat.

track record of significant achievements, right? Of impact with high quality, solving really hard problems, technical or not. And also irrespective of where you might've been, right? Like you, Natalie, you have a background in art and other crazy things that you've done, right?

Didn't you, wait, didn't you work at like Planet Hollywood or something like that? If I remember correctly.

Oh, good, good. I was actually worked at the Hard Rock Cafe for eight years. Oh yeah, I had my wife go out to the seminar. Those are very close. That was a good memory. And then worked in Michelin and James Beard award-winning restaurants in events.

So yeah, completely non-linear trajectory, but learned to be very scrappy, to build things autonomous.

you know, low ego, this all very much resonates. The the last two things I'm going to say are extremely close to my heart. And I'll tell a story when I interviewed.

The recruiter called me a day or two later, I was so nervous. I'm like, oh, my God, oh, my God, how did interviews go? How did interviews go? And they're like, I don't know, give me give me give me the debrief.

But then she emphasized one thing. And my recruiter was Amanda Tennains. I don't know if you're if you're on the call, Amanda, but you're an amazing recruiter.

Amanda said, you know, Joaquin, one of the things that stood out that people really appreciated was your humility. And I'm like, oh, why do you say that? And like, well, because you're willing to come and take a role that's probably like many levels under.

roles that you could have taken, right? And I was like, okay. At first I'm like, why does that matter? And then it struck me and I'm like, oh, wow, I really want to be here because if there's a place that appreciates that, you know, that means that low ego is something we appreciate. And I've learned that this is something that is extremely important at OpenAI, right? It's like, if you want to come for the mission, if you want to come because of what we're doing, that's the right thing. Don't get hung up on, you know, titles, levels, and all these things.

And last but not least, this is one is my favorite probably, is we look for team players. We look for people who are happy to jump out of their swim lane, who can look left and right, see who needs help, you know, if there's something impactful that I can do that's maybe not, you know, on my, you know, on my task list, just go do it, help others.

succeed, help others grow. Absolutely. And you're reminding me of last week, actually. One of my colleagues, Alex Nawar, he curated the nonprofit jam. And there were thousands of participants in 10 different cities. We have a very small team. My team is actually a team of five in total.

And basically, the way we were able to pull off that nonprofit jam in cities all over the country were volunteers from across OpenAI. So from the recruiting team, from the go-to-market team, researchers, our solutions architects, they flew across the country and helped us host these events so that we could put our tools in the hands of nonprofit leaders.

Just, yeah. What an exciting moment. And it really felt good.

like one of the most beautiful moments I've been a part of in all of OpenAI, and I met all of these new people that I had never actually directly collaborated with at the Dallas non-profit jam. It was super rad. So this stuff happens all the time. We're very lucky.

So, Joaquin, moving on, I think that this is also everything you have to say everybody wants to hear, but now we're going to get into the concrete advice.

So what concrete advice can you give to candidates who are interviewing with OpenAI? What works and what do you think doesn't work? And this is probably what everybody's been waiting for. Thank you for being here.

I think so too. Just to remind everybody, we know this is going to be moving very fast. We are going to collaborate to create like a two-pager and share it afterwards. But if we could just touch on the...

some of the big important ones. Yeah, I'm really going to go down to the ground now, like to touch the earth with bare hands. The things I'm going to say are things that I know you all know. But maybe it's nice to have a quick sort of recap. And thank you to the team, again, for providing those. I've done my best to capture them in a summarized form. So if you're a candidate who's interviewing with OpenAI, what are the things you should do to increase your chances of success?

OK, what are the things that you should not do?

Things to do, it starts with communication, right? Make sure that your communication is clear, succinct, and intentional. It often happens, and it's happened to me as a candidate. You're in an interview.

you get a question and you haven't even finished like listening to the question, like you don't even let the interviewer finish their question, you're already in your mind, you're formulating the answer, right? And then often you don't answer the question that was asked, right? That's not good. So it's very important to take a breath, right? Really understand the question. If you don't, ask.

One misconception is people think an interviewer is there to grill you. They're not, right? Take a collaborative approach, right? Think almost like, hey, we're working together here, right? Ask for clarification.

And then second one, then when you talk, when you're going to communicate, be very intentional, right? Take a second to ask what message do I want to convey? And then there's a trick I learned, easy to remember, which is think, edit, speak. So do that. Think, edit, speak.

The other thing I'd say is, show your work. Very often, it doesn't really matter if you get to the correct answer, because we don't really care about that, honestly. Like, what we care about is, are you able to explain your reasoning, and can you convince us that you would have gotten to the right question, maybe with a bit more time? So show your work, show your thought process, show that you know how to break down the problem into its components, and even kind of think of solving the problem together with the interviewer. Don't go solo, you know? You can even sort of validate your thinking, ask for additional assumptions if you need them.

All right, next one. Next one is so obvious, but often the best people, the best candidates, get this wrong.

wrong and fail interviews, prepare. Prepare. Prepare. Practice your interviews. Use chat GPT in voice mode to practice. If you think you're going to be asked about key examples in your career, take time to think about them. Write them down. You don't want to waste time trying to remember.

Give me an example where you led a project that was impactful but complicated, blah, blah, blah. Prepare. Make sure you understand what impact you made in that project. Make sure you understand what impact the project made on the company. And then, of course, understand how your work, your past work, intersects with our mission and what we're trying to do.

The next one up, and this is the last one I'll say in terms of things that work well, is show your curiosity. Don't waste time in an interview asking logistical questions.

questions like comp or level or location or what times you guys come to the office and stuff like that, there'll be time to talk about that later, right? Just be there and focus on the most important things, which are understand the culture, the mission. What impact are we trying to have? What challenges are we facing? And then maybe show your curiosity through past examples in your career. When did you do something which in retrospect seemed crazy, but you were just curious?

And then on things to avoid, look, I have a lot of respect for career advancement. People should care about their titles and their levels and their compensation, but I think there's a time and a place for that. And I think that you don't want to show up leading with that, right? You don't want to show up saying, oh, you know, I'm a VP, whatever, somewhere, and then I want to kind of like be that.

level or something like that. That's like missing the big picture, missing the point. The point is like, hang on, we're building a rocket here, right? Get on it.

And then when you talk about past success, don't use the wrong indicators. Don't go talking about like, oh, you know, I was titled this, I was VP of this, or my team size was that, or like, you know, the scope of my org was that. I mean, that's fine. But the thing is like, what did you do? What did you ship? What impact did you make through your work and through the work of your team? That's kind of like this.

I think we'll probably get a slightly more comprehensive cheat sheet. And I don't think this is really specific to OpenAI. I think this applies to whatever you might interview, right? And I apologize if I'm saying things that are very obvious, but sometimes we tend to forget.

I do not feel like any of this is very obvious. I wish that I would have had this guidance 10 years ago.

The way that I practiced for my interviews, Joaquin, when I was trying to break into tech for a couple of years, I probably went on a 100 interviews. The interviews were my practice. I would have loved to have had this mentorship. I'm completely committed to following up with a list of all of the awesome advice that your team put together for this community. But I think you hit on some really key ones.

And also, sorry to interrupt you, Natalie, but given I anticipate that now at Q&A, we're probably only gonna be able to address a small subset of the questions people might have. There's gonna be a way we can actually turn this into a bit of a rhythm where maybe we can bring some of the people on the recruiting team. If you're listening, be warned. I think this is a...

This is a reveal. We might have some of you be in future sessions and maybe we can focus on some of the questions that we didn't get to answer today. I love that. And for the people in the audience, now you know, this is how we do it at OpenAI sometimes. Sometimes we're like, we have a really good idea. Let's do it. Let's go do it.

Coming in the future of the forum is a series with Joaquin's team and I cannot wait to collaborate with you guys because you're brilliant and kind humans.

Maybe it is time for Q&A, but I do want to ask you one more question, which is what was your most memorable moment so far, Joaquin, before we jump into the audience Q&A? What has been your most memorable moment here at OpenAI?

I had a lot of... I'm cheating because Natalie asked me that question earlier and I was supposed to prepare and I'm like, I can't really choose, there's too many. But I will choose. And this one obviously has some recency bias. This is more tied to the current team, right? So that doesn't mean that I didn't have equally memorable moments, but the Q&A, we had... The all hands, apologies.

The all hands we had with the team last week where the entire recruiting organization got together and I was able to talk to the team and tell them what I had heard in that LGBT voice survey. The emotional connection to the team felt incredibly rewarding.

And I know that that means it's day zero, because now we have to address everything that I heard. But to me, that means the world, that connection, that emotional connection with the team. Thank you, Joaquin. Thank you for sharing that. I hope that your team is on the call right now, hearing all of this. Okay. It's okay if they're busy doing work too. You know, we have a lot of work to do. Just kidding.

True, but sometimes slow down to speed up and I love it when you guys show up to the forum. So I hope they're here. Let's jump into the audience Q&A. We have a ton of questions. Thank you to the team for queuing these up for us.

So Mary Moheby, Program Director at the University of Florida asks, what types of opportunities may become available at OpenAI for experts in the areas of social sciences?

urban planning, community engagement? We have like the crazy challenge of building for the entire world, right? And although we tend to be a bit US-centric, because we're based here.

But when you look at where are people using ChildGPT, a vast amount of people are using it all over the world. And even within the US, what certain communities need is very different from what other communities need.

So I can imagine that as the number of users keeps expanding and the breadth.

of products we build keeps us expanding, we're gonna keep needing very strong user researchers.

We have a role called model designer, design how the model should behave. How do you adapt to certain cultures and needs?

It's hard to sort of spell out specific roles, but I think in general, making sure that we're really building for everyone is gonna need people with that profile.

Absolutely, and just to add to that, Joaquin, there are a few really important communities at OpenAI, not only the forum.

So community engagement is a big piece of what we do because it's part of building AGI for everyone.

And there are so many social scientists that work at OpenAI and policy research, products policy on my team in global.

affairs. So I do believe that there's so much space in the near future. But going back to what you've mentioned previously, we all have to be adopting the tools. So I'm not a technologist, but I had to learn how to use ChatGPT. You have to keep up.

But I just want to encourage the audience, the folks that are not technologists here, that there's a space for everyone in AI from all different backgrounds. I think you, as you mentioned, Joaquin, you just have to kind of also be okay with being afraid. And as you demonstrated in your internship, be okay with not being an expert right away and entering into a landscape that's not super comfortable.

Okay. Another question. Oh my gosh, this is a good one. Sanjush, she's the lead academic data scientist at Kennesaw State University. How does OpenAI design hiring to attract quietly brilliant non-traditional folks who may not apply on their own or be active in the AI space?

How do you find those hidden gems? Yeah, it's a combination of approaches, really. On the one hand, if you're lucky enough to know someone who works at OpenAI or know someone who knows someone who works at OpenAI, our referrals often include people that are extremely non-traditional, but who demonstrate some of those qualities that I listed earlier.

The second one is...

is obviously sourcing. So we have, we divide recruiting into sourcing, which is going and finding the best talent out there, whether or not it's tied to a specific job opening, right? And then we have recruiting, which is making sure that we have the right way to connect people to the roles and manage the interview process the right way and then sort of take the candidate all the way to starting at OpenAI and being onboarded.

I mean, obviously there needs to be some signals somewhere, right? Even for the quiet, silent people, there has to be some output somewhere that we can find, whether it is things that you write or things that you build, right?

code that is in GitHub. So my advice would be do things and find a way to put these things out there so that we can find you.

Yeah, definitely. That makes sense. And a concrete example, Joaquin, because I made this transition, as I mentioned to you, and you know about my background from non-technical into the AI space. And after retooling and getting a PMP and data analytics degree, I did a project where I reached out to all the nonprofits where I had relationships and I built my portfolio pro bono, you know, to have. So I had this output that you're talking about, because that is one thing in tech that I found different than in the nonprofit space or in the restaurant space was you had.

You have to be able to demonstrate that you've done it, that you've built something, so beautiful. I would also say, engage in community events like this one, of course.

And we have, as part of recruiting, I don't know if Celina Mai is on the call. She runs our recruiting programs. And so there's all kinds of events to get maybe non-traditional communities who want to take perspective together and get to know people. Absolutely. I totally agree. Building your network and what we call social capital, you just never know when.

Well, for instance, when I approached you, Joaquin, you were an intern. And when we started planning this event, I had no, I did not anticipate that 11,000 people would be here. But you just never know. You nurture relationships, and you don't know where they're going to go.

take you and what opportunities will surface because of it.

We have an awesome community member here tonight, Halas Nasuli, the head of technology solutions for the US federal government.

And they ask, in a post-AGI world, how are we rethinking the definition of talent? Oh, yeah. That is an excellent question.

The most honest answer I have is I don't know. But now I'm going to share some thoughts. Anyway.

So two lines of thought. The first one is there's a few labor economists out there that are breaking down today.

phase jobs into the tasks that make up that job, and also the skills you need to perform them, right? And also, you know, there's a debate on, like, what does it mean to reach AGI, right? So I think it's important to have that conversation, even as we're, like, on the gradual path to AGI. I think some people argue, well, it's not like one day we don't have it and the next we do, right? So that conversation is going on already.

I think people know about Eric Brynjolfsson and Daron Acemoglu and others, and of course, our own friend of the forum, Karen Kimbrough, and our own Chief Economist, Ronny Chatterjee. And Eric. Eric's been in the forum, too. Eric's always been in the forum, too. Yeah, recently, Joaquin, we hosted AI Economics, and Eric...

Got it.

It was such an honor to host him and teach us. So clearly, things that we're seeing are that there's not only the more mechanical repetitive tasks, like you think, oh, professions like maybe like a paralegal, a bunch of the work would be automated, but also tasks. We're seeing tasks that actually require deep, deep knowledge. We're seeing AI become really good at science in many ways.

And so the things that I think remain very, very human maybe fall into two categories. One is communication, collaboration, leadership, conflict resolution. At the end of the day, with or without AGI, we're still a bunch of humans.

you know, we were designed by nature to be social animals, right? And so, if we are still to work and coexist as a group in a work environment, these skills are going to become, like, increasingly important.

And then the second one is an observation that I've had looking at how so-called AI native developers do their work. It's incredible. It's like their own own technical program manager, in many ways. I've seen people have, like, many codex agents and cursor windows open and have, like, a very disciplined project plan where, like, they stage the work and the project into chunks.

They know how to commit to GitHub intermediate stages with the right amount of documentation. They ask the model to explain, you know,

why did you do what you did, commit so that if something goes wrong, you can actually revert back to an earlier commit, right? And when you look at that, you're like, oh, wow, you're just like a hybrid of like an engineering manager or like a technical program manager here, right? So I think there's something there, right, into managing the work that these agents will do.

And then last but not least, I think there's, at the end of the day, the question of the why and the critical thinking, right, like making sure that whatever work we're doing, whether it's an AGI or a human or hybrid, you know, doing it, is it work worth doing? Is the, you know, are there flaws in the work, right? That sort of a critical reasoning, I think, is going to remain extremely important.

You need to know if the work the AI did was good or not.

Yeah, and if you want to dive deeper into that topic for everyone here, we have discussed the future of work quite a bit. David Autor's come on and he talked about expertise in the work of the future. And he kind of walked us through trends in these very historical significant moments when a new technology was presented and then there were these shifts in work.

And Ronnie and Karen Kimbrough, your former colleague from LinkedIn were just in the forum. So folks can watch the replay and there's nobody on the planet better than Karen who has all of the LinkedIn data to make some of these like share some of these hypotheses with us. But yeah, please everybody just stay tuned because we're very interested in this topic at OpenAI what the future of work looks like for all of us.

So let's keep that conversation going, Joaquin. We'll have you back. Maybe we'll pair you with Karen next time. Oh, that'd be fun.
That'd be fun. Did you know that Karen and I used to be desk neighbors at- I didn't know that. Oh yeah, and we talked a lot.
Awesome. That would make good chemistry for an event.
We have time for two more questions.
So let me choose wisely.
Dheeraj, the Assistant Medical Director at Stanford University School of Medicine asks, how do you prioritize different domains when hiring? For instance, healthcare versus finance versus engineering versus business development.
Yeah, it changes all the time.
That goes back to the question we were discussing earlier when you asked me, how does it feel to be on the team?

CG Simo coming and you all might have read her little manifesto, her thoughts on, I think it's like six big areas of massive impact. I didn't mean little, I meant short, it's not a long essay, but it's super powerful. One of them is, I think the first one is healthcare, right?

So look, I don't know what the roadmap is going to be, but it's very likely that we're going to invest more in healthcare, especially if you saw the health bench results, both like the, I guess like the benchmark to assess how good models are at real world healthcare situations. And also more recently, the study we had with Penda, which is a healthcare provider in Kenya, in Nairobi, I think it's about 40.

thousand actual patient visits and doctors being actually assisted with AI and a hundred percent of them found like a hundred percent of the visits were actually positively impacted by AI. Clearly stuff is happening but to your point look if you think about any domain there's obviously progress and promise in each of them so it's very difficult for me to give you a great answer here.

We do consider them all and I think that there's a balance between trying to do everything and doing fewer things well. What I will say is that I don't think we exclude, even if we're not going all out in a particular domain, I don't think we exclude hiring people as I said earlier from that domain.

If it seems like they're very flexible and open to working on adjacent things.

things, and sort of being on board for the journey. Because one day, one year later, we might very well actually prioritize an area and go all in. Thank you, Joaquin. And I think that that just answered several of the questions. So I'm going to move on to a really fun one, and also honor your humanity. You're not only a technologist. You're not only the head of recruiting at OpenAI. You're a person, Joaquin. Am I? I could be, yeah.

Drew Katowica would like to know, what music and art inspires you in your spare time? Oh, God. Oh, OK, I love it. Mostly music. I'm surrounded by guitars. So I have here, whoops, I bumped something. I have one guitar here. I have another guitar behind the screen. I like to sing folk songs. I sing.

a lot of old Bob Dylan songs, but I feel like my, and I love the lyrics, I was maybe one of the people who was hoping Bob Dylan would actually get the Nobel Prize in literature, so I was actually pretty happy when that happened, because I've followed his lyrics and poetry for many, many years, and I try to sing them too, but my music taste is extremely eclectic, I sometimes listen to Baroque music as well, I like heavy metal quite a bit, all kinds, I think there's maybe very few genres I don't like, there's certain type of techno music that I haven't really gotten into yet, but outside that I think many genres,

one key thing about me, I like to listen to full albums, I don't like to listen to greatest hits here or there, I love when an album is put together and you sort of get into it, like Pink Floyd's

outside of the moon or something. Where it's a whole story. It's a whole story, absolutely.

Well, Joaquin, do you wanna play a little guitar for us? Maybe the Bob Dylan?

Oh, that's crazy. I don't even know if the guitar is tuned.

You definitely don't have to, but it could be kind of cool.

Want me to play with an out of tune guitar here?

Sure, let's try. Let's try.

Oh God. Okay. I promise this is not prepared. In fact, I'm gonna move the camera a little bit down. There we go. There we go.

This is an obscure Bob Dylan song called the Lonesome Death of Hattie Carroll. And it's a very, very heavy song. It's a civil rights song.

So I won't sing the whole thing. Just a little bit. Okay.

Williams and Zinger killed poor Hattie Carroll With a cane that he twirled on his diamond-ringed finger At a Baltimore Hotel society gathering And the cops were called in, and his weapon took from him As they drove him in custody down to the station And both Williams and Zinger for a first-degree murder

Oh, but you, who philosophizes grace Criticize all fears Bury the wretch deep in your face Now ain't the time for your tears

Oh, that was beautiful, Joaquin, thank you so much. And at that, we are at the perfect conclusion. And I can't believe we did this, but it's exactly the right time we were supposed to end. What, like we, this flow is beautiful. And I promise almost all of this was prepared. Absolutely not the guitar piece. I even have to play very quickly, what song do I play?

That was beautiful. And I played one of the super obscure ones. I could have played the Time's Ever Changing or something like that, but that was the first one that sort of came, not to my brain, but just like to my hands. It was perfect. And I hope that performance lives on forever, Joaquin, in the forum.

Well, Joaquin, hopefully this is the beginning and not the end of our collaboration. You and the rest of the recruiting team are always welcome in the forum. I know that the world wants to hear.

from you. And it's absolutely my pleasure to give you the stage to share all the insights. And we're going to move into the one-on-one matching after this. So I just want to say Joaquin, thank you so much for being here.

This is really fun. This didn't feel like work at all. Thank you so much for sharing all of yourself and your experience with such a low ego. It's really a pleasure and an honor to host you.

Thank you everybody for joining and hopefully this was a bit useful and we'll keep engaged.