AIMinds #044 | Aaron Wang, Co-Founder & CEO at Apriora

About this episode

Aaron Wang, Co-founder and CEO of Apriora. Aaron Wang shares his journey from academia and AI research at Facebook to founding Apriora, a company transforming hiring through autonomous AI recruiters. He discusses the inefficiencies of traditional recruitment, the rise of agentic AI, and Apriora’s innovative solutions.

Listen to the episode on Spotify, Apple Podcast, Podcast addicts, Castbox. You can also watch this episode on YouTube.

In this episode of AIMinds, hosts Demetrios and Aaron Wang interview Aaron Wang, CEO and co-founder of Apriora, an AI-driven recruitment company. Aaron discusses his academic background in computational biology, applied math, and economics at Brown University, and his work at Facebook’s research labs on projects like the Orion glasses and self-supervision techniques.

Aaron also explains how his experiences led to the creation of Apriora, an AI-driven company that automates key aspects of the recruitment process. He highlights how AI improves efficiency and decision-making from candidate screening to hiring, benefiting both recruiters and applicants. The discussion also explores AI’s broader impact on the job market and the evolution of employment technologies.

The episode explores AI's impact on business processes and concludes with insights into the challenges and opportunities of integrating AI into traditional industries like staffing and recruitment, highlighting the strategic use of AI to create value in the labor market.

Fun Fact: Aaron accomplished a remarkable feat by triple majoring at Brown University in Computational Biology, Applied Math, and Economics, which he sees as various forms of applied math based on different datasets.

Show Notes:

00:00 Studied computational biology, applied math, economics.

05:37 Personalized AR using self-supervised computer vision.

09:32 AI's potential in recruiting is now transformative.

12:47 Streamlining hiring phases with AI-generated job descriptions.

15:40 Hiring process evolving with technology and AI.

18:01 Automating recruitment for large corporations' efficiency.

More Quotes from Aaron Wang:

Transcript:

Demetrios:

Welcome back to the AI Minds Podcast. This is a podcast where we explore the companies of tomorrow being built AI First. I am your host, Demetrios. And this episode, like every other episode, is brought to you by Deepgram, the number one voice API on the Internet today, trusted by the world's top enterprises, conversational AI leaders and startups. Some of them you may have heard of, like Spotify, Twilio, NASA, and Citibank. Today we're joined by none other than Aaron, the CEO and co founder of Apriora. How you doing today, man?

Aaron Wang:

Hey, man, great to see you, Demetrios, as always, and excited to have a fun chat. Should be cool.

Demetrios:

Okay, so I gotta start with this. You triple majored at Brown. How did you survive triple majoring? Did you just stay inside all day, every day? That is so wild. I have only heard of these mythical creatures in fantasy books.

Aaron Wang:

Well, I don't know what, what fantasy books your parents were making you read, but no, it was a super fun time. I, I, I really just wanted to learn as much as I could. So for context, I was studying computational biology, applied math, and economics. And I all thought of, I really thought of all of them as kind of different versions of applied math. Like, obviously, applied math is applied math, and that's reflexive, but, you know, computational biology, it's just, it's just a different data set. Are you looking at, you know, the labor market data or are you looking at, you know, genetic data? Right. Um, and so I found that, I found that super fun. And also, you know, did my master's there while I was there.

Aaron Wang:

And yeah, it was, it was a great time. I was endorsing. Incredible.

Demetrios:

Yeah, it seems like you got, you squeezed every last drop from the juice that you could get. So I like it. And now after that, you went off and you became a researcher at Facebook, and one of the things that you worked on was these glasses that just came out.

Aaron Wang:

Yeah, those Orion glasses. Yeah. So I'd spent some time at a fair, which is. Yeah, that was incredible. I was in Menlo Park. I guess my main project was on self supervision, which is particularly something called the Barlow twins technique, which was invented at fair. And it's this idea of how can I quantify the similarity between different frames within a video? Because if I can do that, then I can do. I have a lot of interesting downstream tasks.

Aaron Wang:

So, for example, if I'm wearing those Orion glasses, I could say, oh, man, I lost my keys. Like, where are my keys? And then what I can do is I can Then scan all of the previous video recordings, look at every frame, and say, okay, what does this look like? There are a pair of keys in here. And then based on that, I could tell you, okay, where the keys are. And then, like, based on the GPS location, you know, help you track your steps back to exactly where those keys were. So that's a bit about what I worked on there. Super fun.

Demetrios:

I needed this last week when I was in Amsterdam and I forgot my backpack at somewhere. Uh, I was just riding my bike home or back to the hotel, and then I thought, wow, it's a lot easier to ride the bike for some reason right now. And I realized at that moment I did not have my backpack on me with the laptop weighing me down. And so I instantly turned around and I had gone to eat. And then I went for ice cream after that. So I didn't know if it was at the ice cream spot or if it was at the restaurant. But luckily enough, it was tucked away right where I left it when I.

Aaron Wang:

Got back to the restaurant.

Demetrios:

And the.

Aaron Wang:

Yeah, the laptop was there. Okay.

Demetrios:

Yeah.

Aaron Wang:

Yeah. Biking culture is crazy in Amsterdam. That's awesome. Yeah. I always feel like there are a lot of cyclists in sf, but Amsterdam's. Amsterdam's different, dude.

Demetrios:

One of my favorite things to do, I realize, is ride a bicycle in Amsterdam. This is a very far tangent from all of this research work that you were doing at Facebook. But honestly, like, I am so happy on a bike in Amsterdam, and that is. Yeah, it's my happy place. As I get older, I recognize it.

Aaron Wang:

Yeah, it's. It clears the mind. And, you know, you see everybody else on bikes as well. It just feels like super communal.

Demetrios:

Yeah, it's old architecture that.

Aaron Wang:

Yeah, I mean, we could. We could talk forever about that or the architecture in Amsterdam. It's. It's really. It's really one of a kind.

Demetrios:

Yes. So, okay, getting back on track, because that's my one job for this interview. I would love to know a bit more about the stuff that you were working on with these glasses. Did it have to do with the segment. Anything model that came out? Is there any overlap there?

Aaron Wang:

Yeah, we. No. So that was a. That was a different model. Those were. Some of the other folks were working on. Um, okay, what we were working on was something called. I think it was called Pixar, and I'm trying to remember exactly what it was, what the acronym was for.

Aaron Wang:

It was something about personalized experiences in augmented reality, something like that. And the whole idea was, you know, how can we use Kind of the state, the latest in computer vision and particularly self supervision which was getting big at the time, you know, running it on these like big clunky glasses. There's like a lab in Fremont where we were just walking. You know, you're walking around, it's like the set of like a perfect home and you're trying to like figure out like, okay, like I gave it a picture of a fire extinguisher, right? The glasses, I give it a picture of the fire extinguisher as I walk around. Like if I see a fire extinguisher in my kind of periphery, does it give like a high score? Right? So that was really interesting. So if I was like looking at the dining table, it'd be like, no fire extinguisher here. Turn my head to left, boom. Fire extinguisher 100%.

Aaron Wang:

It's right there. So a lot of object detection point cloud work slam. Familiar with that.

Demetrios:

So the fascinating part about all of this to me is that this was years ago, right?

Aaron Wang:

That's right.

Demetrios:

It took this long for it to be shipped into the glasses that are finally this consumer facing product.

Aaron Wang:

It takes a long time. I think. You know, Mark Zuckerberg made a really great bet on open source and on these large language models and he's also made huge, he made really big bets in hardware as well. I mean Facebook has one of the largest computing centers on the planet and you don't, you know, you wouldn't think about it like, like that or you wouldn't think that they had been kind of collecting all this hardware, you know, a couple of years ago. So yeah, they, they, they work, they, they think about this stuff in the long term. And I, I, I think fair is, is one of the best labs in AI today.

Demetrios:

So how did you change and what inspired you to get into more of a voice or AI agents type of work?

Aaron Wang:

Yeah, I'm not tied to a particular medium for, for AI, I want what's most natural. So for a glasses medium, computer vision is, is the most intuitive and the most natural. I'm probably not going to be talking to my glasses all day. Um, and so, but the glasses will be, you know, seeing what I'm seeing. Right. And so computer vision is very natural. But there are other instances. You know, I worked at, I worked in even before fers at doing natural language for co generation which believe it or not, you know, that was actually pretty difficult back then.

Aaron Wang:

You'd have a whole separate model just to do, you know, text to SQL. And now, you know, One big LLM can just do all of it. Yeah. But for that, if we're doing cogeneration, obviously I'm going to want something a little bit more kind of NLP based. Right, that's obvious and probably isn't very interesting. But voice AI is particularly interesting as a medium today because it's really one of the highest bandwidth, kind of channels of information that humans use today. We're chatting today and people are listening to us on this recording. Um, and so what's most exciting is that technology is just coming to fruition today, you know, with companies like, like Deepgram and, and others that are allowing for essentially instantaneous inference, which is, you know, we haven't seen this before.

Demetrios:

And what was the inspiration behind the becoming an entrepreneur? Going out and starting a product like Apriora.

Aaron Wang:

My last job before this I was working in at a quant hedge fund and it's great. I think if you want to work any job, that's the place you're going to want to do it. I think at some point you got to understand that there's a big why now? Moment in AI and the past decade in SaaS, it's been wandering around and a lot of the kind of interesting products have been created already. But now with this platform shift, there's a lot more that you can build that just hasn't been available before. A lot more you can do to serve the customer in a way that you haven't been able to before. I think we're in the kind of hiring space or the recruiting space which historically hasn't been particularly attractive for a whole multitude of reasons, but really because of value that you can create for your customer. And that's completely changed now. You can actually do make incredible improvements for your, for your customers that through voice AI and through, through other types of AI that you just really haven't seen.

Aaron Wang:

So that's why, you know, we decided to take the jump.

Demetrios:

And what does the product look like? Because as you mentioned there, it's not typically seen as the sexy space, but there is a lot of use cases that you can have and plug AI into when it comes to hiring and onboarding a new team member. And everyone does it.

Aaron Wang:

Yeah, yeah. Well, I can give you an example of just how kind of interesting the space is. So staffing agencies, right, they're like services businesses. How many public, you know, Demetrios, just throw me a number here. How many public staffing industries just in the U.S. do you think there are like, you know, IPO'd, you know, on.

Demetrios:

The New York Stock Exchange, maybe 20.

Aaron Wang:

Yeah, they're. They're over 40, right. Which is pretty incredible when you think of the size of these things. And you don't really think about that, right? Like, whoa, these are services businesses. You know, you're telling me these are, you know, it's historically not the most successful. And, I mean, you'd probably be right. You know, the gross margins are really, really tight. Right.

Aaron Wang:

For every recruiter, I can only hire so many people. And so what we're doing is very interesting where, hey, look, if you can actually replace or do a better job of 80 to 90% of these tasks that these recruiters are doing completely autonomously, right. Like an autonomous agent, then that's a huge lift in your gross margins and the kind of product that you could deliver, both in terms of quality and the speed at which you can get to the candidate. Schedule time with the candidate, interview the candidate. If they're great, like, let the employer know right away, all right, these are huge, huge kind of value accretion points that you haven't seen before, before agentic AI.

Demetrios:

So I like how you break down different phases that it feels like we all go through when we're trying to bring on or we're hiring someone new and you want. If it's just me going out there and saying, all right, I need a head of customer success or I need a head of marketing, I'm going to potentially put out a job description, which may take me a little while to create these days. It's easier because I can just ask ChatGPT or something, say, give me a job description. Here's some important key factors that I want to keep in mind. And then I throw that out there and I get flooded with applicants, ideally. And then I have to sift through them, and then I have to do that first interview and that's time. And then I want to introduce them or socialize them to my team, see if they're a cultural fit, do second interviews, third interviews. So what are you doing to flip that all on its head?

Aaron Wang:

Everything that you mentioned you can now do autonomously. So someone applies to work at Deepgram. You know, they're not going to. That applicant isn't going to wait a week for the recruiter to review their resume and, you know, send out a batch of emails to all the people that they shortlist. Right. The second that you apply, you're going to get an email or an SMS from the AI recruiter and it's going to say, hey, look, we reviewed your Application. You look great. Find some time on my calendar.

Aaron Wang:

Here's my calendar link. Let's chat. You can chat with her tonight at, you know, 11pm you can chat with her now. Schedule some time. I hop on a zoom call or a phone call, let's say, you know, an interview of a video interview, like a zoom. You hop on, you have a 20 minute conversation, you know, technical, non technical, you know, the whole nine yards. And after that she'll send you, hey, thanks for taking the time. We'll be in touch with you with the next steps.

Aaron Wang:

Write the feedback directly for the company and the AI will suggest, hey, this person, Demetrios was really, really great. He's got great communication skills. His Python understanding is incredible. He talked about his experience at Facebook. Really great candidate. I think you should hire him. And now it's up to the recruiter. They can just send them straight to the hiring the candidate straight to the hiring manager, for example, a cultural fit test.

Aaron Wang:

But other than that, they're a great fit.

Demetrios:

You undoubtedly, because I imagine you're steeped in this space, saw that video that got pretty viral a few weeks ago where it was a candidate's agent talking to a company's agent. How do you feel about those types of interactions? Do you think that's going to be something that is a bit more common as we move forward?

Aaron Wang:

You know, I talked, I know we talked a bit about, you know, I spent some time in the hedge fund space. It's funny, I do think of hiring almost as a market in the sense that, you know, applicants are going to get better and better and better and smarter and smarter and smarter, use better technology. And thus, you know, employers are forced to do the same. Right. So everybody's Kitty's candidates are creating AI generated resumes and submitting, you know, 200 job postings every day. If I'm a recruiter, I just don't have time to look through all these resumes. And so I need to leverage technology to be able to, you know, tell me, hey, I've got 100 people here who are the top five that I actually need to chat with. Right.

Aaron Wang:

And so in any market there's some kind of adversarial nature and you want to make use of technology. I think this will only continue. I think for employers it's going to be really important to detect if they're using some sort of AI tools. So we have our own cheat detection system which looks at, you know, their video, it listens to their keyboard. You know, are they tapping on the Side. Are they, you know, do I. Am I seeing them look at another screen and say, hey, look. Two minutes into the video, I can hear tapping on the keyword.

Aaron Wang:

Every time I ask them a question, they're tapping on the keyboard. The AI recruiter will. Recruiter will write up those notes and share them with the. With the team. And so it's going to be a cat and mouse game. But, you know, our hope is to keep the employers ahead of the curve.

Demetrios:

Now, you mentioned to me before we hit record, that one thing that was important as you went through YC was finding the vertical and finding who you were. Your ICP is. Can you explain a bit about that process?

Aaron Wang:

Yeah. So historically, um, hiring technology or recruiting technology has never been a question about how big the market is. The market's huge. Every company hires. Um, the question has been about how much value can you create for the customer, Right. If I. If I go and sell an ATS or kind of an assessment platform, right, how much time am I really saving and how much money am I actually saving our customers? Right? And turns out it ends up not being a lot in with previous technologies. But here we have something that does create a lot of value.

Aaron Wang:

You know, automating, you know, again, the, you know, 70 to 90% of what a recruiter can do today. And so the question then becomes, okay, this market's really big. Where do we start? Right? And so for us, we tried a bunch of different ICPs, and we found that for us, we really focus on large corporations is a big part of it. Of course, they see a lot of volume and require. I got these 40,000 applicants that applied for this graduate recruitment role. I just want to know who the top 20 are.

Demetrios:

Excellent, dude. Well, this has been awesome talking to you. I appreciate you coming on here.