From Harvard at 18 to Building Crypto's Fastest Exchange | Vlad, Lighter

What does it take to build an exchange that refuses to choose between speed, cost, security, verifiability and composability? Lighter founder Vlad Novakovski joins David to trace his path from international Olympiads and Harvard at 18 to Citadel, early machine learning, Lunchclub and an 18-month ZK engineering bet. --- 📣SPOTIFY PREMIUM RSS FEED | USE CODE: SPOTIFY24 https://bankless.cc/spotify-premium --- BANKLESS SPONSOR TOOLS: 🔓NEAR | TRADE CONFIDENTIALLY, GET 20% BACK https://bankless.cc/near-20 📊BITGET | TOKENIZED STOCKS 2.0 https://bankless.cc/bitget-stocks 🧭OKX | TRADE, EARN, PAY to OKX | 120M+ USERS WORLDWIDE https://app.okx.com/join/USBANKLESS 🎯THE DEFI REPORT | ONCHAIN INSIGHTS https://thedefireport.io/bankless 👑BANKLESS PREMIUM | AD-FREE & BONUS EPISODES https://bankless.cc/spotify-premium --- TIMESTAMPS 0:00 A Particular Set of Skills 11:33 Harvard at 18 19:03 Citadel and the Search for Alpha 25:14 Machine Learning Before the AI Boom 36:46 ​​Finding Elite Talent Early 41:39 The Lunchclub Matching Problem 45:11 Pivoting Into Lighter 51:18 The 18-Month ZK Engineering Bet 56:38 Lighter’s Five Pillars 1:00:04 A New Generation of Exchanges --- RESOURCES Vlad Novakovski https://x.com/vnovakovski --- Not financial or tax advice. See our investment disclosures here: https://www.bankless.com/disclosures

• 0:03Speaker 00

  Hey, Banglais Nation. In this episode, we're going to talk to Vlad Nowakowski from Lighter. Vlad is a pretty interesting character, and he's been around the podcast circuit quite a lot, talking about Lighter as a platform, how it's different from hyperliquid and other perp platforms, and what his goals are. But I don't think too many people have really explored the beginnings of Vlad, going all the way back to his teenage years when he was just a math Olympiad, competing with alongside some of the brightest minds of our age, including Dario from Anthropic, and Vlad from Robinhood, and a bunch of other names that have gone on to raise massive companies. And so, the who Vlad is, I think, is a pretty unique and interesting character. And I just really wanted to ask and answer the question, why do investors love Vlad? And so, this is what you're going to hear,

• 0:46Speaker 00

  a different side of Vlad from Lighter. So let's go ahead and get right into it. We record this in person in a studio in Manhattan, and the beginning of this episode, we just kind of start rolling. So maybe it feels like you started the middle of the conversation, because you kind of do. So let's go hear from him right now. Vlad, I want to know a little bit more about your background, because people know you as the founder of Lighter, the Perp Dex, theorem layer two, Perp Dex. But your background goes pretty far, pretty far back into your childhood, I think, with your first relevant set of skills. Talk to me about-

• 1:15Speaker 01

  Particular set of skills, right?

• 1:17Speaker 00

  Particular set of skills, yeah. Talk to me about math. You won a math and physics Olympiad. Me and most of my listeners probably don't know what that means. What does that mean?

• 1:28Speaker 01

  Yeah. Well, I started competing in, you know, academic competitions, let's just call it that, when I was around 12. And I guess some of the highlights of that were making the US team to the International Physics Olympiad, and the International Olympiad of Informatics. On the math side-

• 1:45Speaker 00

  Informatics?

• 1:46Speaker 01

  Yeah, that's the programming one.

• 1:47Speaker 00

  Okay, programming.

• 1:48Speaker 01

  The math, I mean, I tried to do a hat trick as far as making all three. I was actually the first, or one of the, I think, one of the first people in the US ever to make two teams. Didn't quite make three, but, you know, but it was good.

• 2:02Speaker 00

  What does it mean to compete in math and physics? Like, how does one actually, what is to compete over?

• 2:09Speaker 01

  Well, so the Olympiads all have a different format. The programming one is maybe the most, would be like the most exciting to watch. There's actually are like, like these days, like Twitch streamers and whatnot who do that, where it's like you're writing code, basically have a problem solved. Like, I don't know, the problem could be like, here, you know, here's the location of two pizza places in New York, find the shortest path given traffic, how to get from maybe, and you have to actually like, it's not theoretical, you actually have to write a program that would actually do it.

• 2:39Speaker 00

  Right. It's deterministic. So there's a very clear, correct answer?

• 2:42Speaker 01

  Yes. Okay. Yes. And so, like, you have, you know, three problems like that, five hours, two days of that. So that's the, that's the programming one. Math, at the, I'm talking about the highest levels, right? And the ways leading up to that, it's, you know, it's not five hours, right? There's like, it's less intense. But then the math, that one looks like, so it's also two days, and you have three problems where you have to find a solution with proof. You've probably heard of like, the four color theorem, the like, you can like, paint like a map with like four colors, like any like, it doesn't matter if it's states, countries.

• 3:18Speaker 00

  Right.

• 3:18Speaker 01

  Now, that one's very hard, but let's say the problem was like, prove that you can do it with like, six colors.

• 3:22Speaker 00

  Okay.

• 3:22Speaker 01

  Hypothetically, right? And it's like, it's not, I don't know, just to show one example, you have to like, prove that you can always do it. And then you have to write proofs like that for like, three problems over five hours, also, I think, two days. Physics one, there's a theory part and experiment part. So the theory part is also similar. It's like, you know, like a block of ice is coming down the hill, you know, with, you know, with kind of assumptions like, you know, how fast is it gonna get there while it melts, you know? So it's not just like, basic, it's not like Physics 101, where it's like, just like a rock, but like, because it's ice and melting, you have to like, do something, bring in a bunch of different concepts together. But then there's also the experiment part. And that, that's probably the most different of all these. You like,

• 4:04Speaker 01

  you actually have to like, do stuff with your hands and like, you know, measure something or like, build something, design something.

• 4:10Speaker 00

  And this was all fun for you. You, this, I'm assuming you were like, intrinsically just thrilled to do this math, to be in these competitions, right? Like, math is entertaining, entertaining for you?

• 4:21Speaker 01

  Yeah. So the thing was that, like, it was like a virtuous cycle, because I was pretty good at it even before competing. And then, once I started competing, I saw that, like, winning is fun.

• 4:31Speaker 00

  Okay.

• 4:32Speaker 01

  And then, and not only is winning fun, meeting other like-minded people. And in the long run, that's actually the most important part of all this, is the people you meet.

• 4:41Speaker 00

  Right, because this wasn't solo activity, this, you were on a team.

• 4:44Speaker 01

  Well, yes, but even, I mean, a lot of the competitions are individually measured, and then the country, you know, once you get the international level, the countries have teams. But you're just like, you know, because there's training camps, like, as it was, you know, you sit around, wait for results, like, you meet a lot of the people. It's like, you know, the opportunity to meet other like-minded, you know, students, but really just kids around the country.

• 5:09Speaker 00

  Right. Who did you meet?

• 5:11Speaker 01

  It's funny, because a lot of the, I mean, back then, these are people that I just met at the camps. I mean, they were impressive people then, right? Just because of their skills. But some of them now are like, shaping what we see in industry, like, for example, at the US, the training camp for the US Physics Olympiad, two of the co-founders of Anthropic were there. So I got to meet them, got to know one of them quite well, actually. And then at the Intronics one, you know, the first CTO of Facebook was, I mean, they weren't, they didn't-

• 5:39Speaker 00

  The future CTO, yeah.

• 5:40Speaker 01

  Yeah, the future, yeah. So some of the people, you know, who's also on the board of OpenAI now, right? So like, that's pretty interesting. And that's, that's just like, people who've achieved, you know, the kind of the most well-known outcomes, but there are a bunch of others who, right? Like, for example, like, I think, if you look at people who run some of the big quant trading shops, like, you know, you see a lot of people from those competitions there, et cetera.

• 6:06Speaker 00

  People that come out of these like, math and physics and informatics Olympiads, Dizzy, was your cohort unique in that a handful of them grew up to be incredible founders, shaping the future of the world? Or like, is it always like this?

• 6:21Speaker 01

  Yeah. It's a good question. I mean, I think it goes, there are bursts, like, I think the early 2000s, for whatever reason, maybe it's because we were the first to grow up with the internet. That cohort in particular, and not just on the math and physics side, like, that cohort, you know, like, achieved a lot of success. I mean, you know, the vice president is part of that cohort too, right? Like, you know, people who've come, who've done very well in politics, who've done very well in other forms of business, like, that early 2000s, I think, there was a lot of, a lot of great achievement that came out of that. But, but, but that being said, like, if you look back even to the 90s, like, one interesting thing was, the 93 team that Harvard sent to the ICPC, which is like, the equivalent

• 7:03Speaker 01

  of the Informatics Olympiad at the college level, same competition, team of three. So out of that three-person team, not one, but two became billionaires. And like, in the 2000s, not, not now, when 2000s was, that was a very high bar.

• 7:17Speaker 00

  Yeah. I would imagine the competition has plenty to do with it, because like, what is the common denominator between physics, math, and informatics? You know, there's a lot in common there, but really, it's the competition side of things. If you're good, you were, you were almost, as you said, did the hat trick. What kind of problems, whether it was math, physics, or informatics, what kind of problems, what, what type of problems did you really enjoy the most? Is there, is there something to discuss there?

• 7:45Speaker 01

  Yeah, yeah. This was always like, you know, when we would hang out after the competitions and talk, this would always come up, like, it's different, even of the people who did well, we all had like, our own, like, favorites, or like, the kinds of things that, like, we thought we grogged better than others.

• 8:02Speaker 00

  You were particularly strong at?

• 8:04Speaker 01

  Yeah. And this was always a debate, like, some guy would be like, oh, like, I love, like, geometry is really cool, I have geometry. And then, like, maybe somebody else would be like, oh, no, like, that, like, like trigonometry is really cool, right? Et cetera. But like, for me, I really liked the stuff where you needed like, an unusual insight to solve it. Or if you look at the problem one way, it looks like there's like, no solution or very difficult path to solution. And if you look at it like, sideways, like, oh, actually, like, you can solve this in two lines. And, I mean, I didn't always find those, but when I did, I think that's when I learned the most, and that's why I got the most satisfaction out of it.

• 8:38Speaker 00

  This stroke of genius problems.

• 8:40Speaker 01

  Well, I guess, Gene, you know, I think it's more of just.

• 8:44Speaker 00

  Or thinking outside of the box.

• 8:45Speaker 01

  Exactly, exactly.

• 8:46Speaker 00

  Thinking creatively.

• 8:47Speaker 01

  Right, exactly.

• 8:48Speaker 00

  Yeah.

• 8:49Speaker 01

  I mean, you didn't, you know, these problems, solving them, you don't have to be, you know, Einstein. You're not thinking of something that no one's ever thought of before, but you do have to think outside the box.

• 8:57Speaker 00

  I've been trading crypto assets for almost a decade, and I've used so many wallets, exchanges, aggregators, frontends. And I'm kind of always looking for the same thing. One interface with deep liquidity across a bunch of chains and assets and the ability to act in private. And I still control my own funds and I've never really found it. And I'm always switching wallets, juggling gas fees or getting eaten by slippage. Near.com feels fundamentally different to me. My account is easy to use and I can take all the actions I want from any chain while my activity remains confidential. It runs on Near, which has moved over $23 billion cross-chain using post-quantum signing and has over five years of uptime. Near.com is the best way to be on-chain and be in control. Get 20% of your trading fees back through the bankless link. It's in the show notes, not investment advice.

• 9:42Speaker 00

  Markets don't move one asset at a time. One day it's Bitcoin, the next Nvidia, then gold, and then the S and P. But most traders are still managing their portfolio across different platforms, different accounts and different pools of capital. Bitget just changed that. Their new Stocks 2.0 product lets you trade tokenized equities directly with USDT, all inside the same app you already use for crypto. This is not just another tokenized stock product. Stocks 2.0 is designed around deeper liquidity, faster execution and the lowest fees in the market at just 0.04% and one-to-one economic exposure to the underlying stock. Dividends, stock splits and other corporate actions are reflected automatically, helping your position stay aligned with the asset you actually want exposure to. One platform, one account, multiple markets, crypto equities, commodities and more, all accessible with USDT. Bitget, trade smarter.

• 10:27Speaker 00

  Start trading today through the link in the show notes. This is not investment advice. Trading is changing, not gradually, right now. OKX just launched Trading Boss directly inside the OKX app. Grid trading, DCA, arbitrage, you set your strategy once and it executes around the clock. No staring at charts all day, no manual entries, no missing moves while you're asleep. And for the first time, automated trading actually feels simple. But OKX is thinking bigger than just trading. They also launched the Agent Payments Protocol, an open standard that lets AI agents execute full commercial transactions on-chain. The Ethereum Foundation, Uniswap and AWS are already building on it. And now it's live inside the United States. And new users who deposit and trade can get up to $500 in Bitcoin through the bankless link.

• 11:10Speaker 00

  The link is in the show notes to learn more, not investment advice, not available in New York or Texas. Your teenage years were all super heavy in math. And this was also where you also met Robin Hood's lad, correct?

• 11:20Speaker 01

  So we were at the same high schools, called Thomas Jefferson, which is a math and science high school in Northern Virginia. Yeah, we were the two lads. I was, you know, Russian immigrant. He was a Bulgarian immigrant. Our paths crossed more later in life, like, I mean, we, you know, we were at the high school. Of course, I mean, he started doing math actually more deeply in undergrad and then graduate. He was a PhD student under Terry Tao, who was probably like the best mathematician in the world, you know, UCLA. And so, so anyway, so then he got into high-frequency trading later as I did. So we, our paths actually got closer kind of later on.

• 11:57Speaker 00

  Was it obvious to you, being in these Olympiads and just in these cohorts of people, was it kind of just obvious amongst you and your peers that everyone was going to go and apply themselves in some great way? Or were you guys just kids playing with numbers?

• 12:13Speaker 01

  For me, I saw it a little bit because I used to think of it like, okay, if you, if you like stack rank everybody, like, you know, like the people, like, well, let's say you were like, whatever metric you come up with, like you stack rank everyone. Let's say in the country, right? Like whoever the top seed is, second seed, third seed. If you like, and that ranking doesn't change that much from one year to the next. If you, I thought, okay, well, if you project that out 20 years, it's probably going to be not about the same then. So like, these are the people who will contribute. And now I think the thing that was different is like, I think, because these competitions, the people who organize them are academics. So you're, the default career path is academia.

• 12:58Speaker 01

  So I think, like, we weren't, obviously, we weren't thinking about like Facebook and Anthropic and, you know, OpenAI. We were thinking more like, okay, these people are probably going to become like, like Harvard and MIT professors down the road in their field. So how exactly they apply their skills was not something anyone was really thinking about. But, but yeah.

• 13:17Speaker 00

  When did, as you're growing up and just kind of getting into your adult years, did a particular direction for you open up and become clear? Or, or how did you, how did you go from, you know, a kid playing with numbers and competing with other kids into like thinking about a career, thinking about being an entrepreneur?

• 13:34Speaker 01

  Well, one, one thing that I realized at some point around then is like, I really like data. And that's, the thing is that you would think, well, yeah, like numbers, data, of course, but actually the competitions don't really involve data. The only, out of the competitions I described, the only one that does is the experimental physics one, where you're actually dealing with the real world data. Right? And the funny thing about that was like, when I was training, I didn't do that well in the experimental part. I would always like be the first at the training camp in the US on the theory. And then I would like not do that well in the experiment. I would hope that I would still make the top five even with that. But the actual international one, like, I actually had like a pretty bad day on theory, jet lag, whatever, I don't know.

• 14:17Speaker 01

  But in the experimental one, I was ranked second in the world. And so overall, I still got a silver. So that was kind of the start of like, I mean, okay, I actually really like this like data, real world stuff. And that I think led to doing more with applied sciences later on, like economics, finance, trading and kind of so on and so forth.

• 14:35Speaker 00

  You met Ken Griffin sometime in your teenage years, correct?

• 14:38Speaker 01

  Yes, I finished high school at 16 and Harvard at 18. So yeah, I was.

• 14:44Speaker 00

  You finished Harvard at 18?

• 14:46Speaker 01

  Yeah.

• 14:46Speaker 00

  How does one do Harvard in two years?

• 14:49Speaker 01

  Well, two and a half.

• 14:49Speaker 00

  Two and a half years.

• 14:50Speaker 01

  Yeah.

• 14:51Speaker 00

  How does one do that?

• 14:53Speaker 01

  You know, there's like, you can get credit for one year from, because I took a bunch, you know, like, we were competing on a lot of things, right? Including how many APs you took. Like, I'm talking about among that competitive cohort. So like, I mean, there's one kid doing 17 APs. I think I did like 12, but that's enough to get one year of credit at Harvard. And then, and then I just, like, instead of taking five classes a semester, I took six to get another semester.

• 15:16Speaker 00

  Was it about finishing Harvard, like, that was it? Another competition is just like, who can finish Harvard quicker?

• 15:23Speaker 01

  Well, that was more a competition between me and myself. No one else was really trying to do that. The, I mean, I looked at it, like I talked about thinking outside of the box, right? Like, the way I looked at it was, you know, assuming I, now, these days, probably someone like me would have dropped out and gone to like, you know, start, you know, build a protocol or build an AI model or something. But back then, that was rare. Like, the only case we really knew about who did that was Bill Gates, right? So the, I mean, and then, of course, one of us in that cohort did it, you know, with, with Zuck, but like.

• 15:54Speaker 00

  Right, Zuck hadn't done it yet because, right, okay, yeah.

• 15:57Speaker 01

  But like, but anyway, like, so that wasn't really a path I was considering, right? And so it was like, okay, well, like, I might do something in the industry, but if I'm going to stay in school, the way I look at it is like, it's strictly better to be a grad student than an undergrad. Because you get undergrad, you have to pay them, grad student, they pay you. So I was like, I should finish the undergrad as quickly as possible. And then I can, actually you can cross-register. I can literally take the same exact classes as a graduate student and be on the same exact campus. And it's like, I was like, I might as well do that.

• 16:25Speaker 00

  So your form of dropping out was just doing it faster?

• 16:28Speaker 01

  Yeah.

• 16:30Speaker 00

  Why get so aggressive and competitive with the timelines? Why not just do Harvard in four years?

• 16:36Speaker 01

  Because of this thing, right? Because, because it's strictly better to be a grad student than an undergrad.

• 16:40Speaker 00

  So this is, you, this is an optimization problem?

• 16:43Speaker 01

  Let's say I wanted to stay for seven years. I would have spent two and a half as an undergrad and four and a half as a grad student versus the other way around, right?

• 16:49Speaker 00

  Sure, okay.

• 16:50Speaker 01

  It's just strictly better. Because you can cross-register, so you can take the same exact classes. And this is just like a life hack that no one really thought about.

• 16:56Speaker 00

  I mean, so obviously you got an education out of Harvard, but what did you get out of Harvard beyond that?

• 17:00Speaker 01

  Learning how to, you know, like, there was this thing they said, which sounded like cliche, but I think ended up being true, which is like learning to think in different ways. So, you know, it was one thing in like, within a certain field like math, to look at a problem within two different mathematical approaches. Like, we're already doing that, but like, to think about a real world problem from, like, imagine thinking about a real world problem as a mathematician, as a philosopher, as, you know, biologist, you know, like, thinking about problems in completely different ways, I think, was like, on the academic side, what I learned. I think on the personal side, just like met a lot of great people, like, I think I really learned, like, again, this is cliche, but like, how to network?

• 17:42Speaker 01

  Like, how do you like go into a room and like start conversations with different kinds of people, get to know them?

• 17:47Speaker 00

  Kind of in the same way that you were competing in the physics and the math and the informatics Olympiads, but really the common denominator of all of those is competition. At Harvard, you're learning all these different ways to think, but really the common denominator of Harvard is you're hyper-social at Harvard. Like Harvard, Harvard is a super social environment.

• 18:03Speaker 01

  That's right.

• 18:03Speaker 00

  And you need to be in, in that game to really excel, I think, coming out of Harvard.

• 18:08Speaker 01

  Well, it depends on the field, right? If you're like, I mean, I guess, like, most scientific fields have some collaboration, but obviously there are also others like that reward more like the lone wolf archetype too. So it really depends on the field. But yeah, for, for a lot of the fields, the value you get out of Harvard is through like A, collaboration, but B, meeting people from other fields.

• 18:28Speaker 00

  So your first step out of Harvard was working with Ken Griffin?

• 18:31Speaker 01

  Yes, I guess that was the original question. We got sidetracked. So can, you know, like he had, he himself had started Citadel out of Harvard. I believe he did, he graduated, he wasn't a dropout. I think he graduated, but he was running the fund out of his dorm room in like 1988, 1989, like 15 years before I was there. And so I think he, once he got to know my background, I mean, I first interviewed with the firm like everybody else, but then, you know, they kind of, you know, he got to know my background and spent a lot of time and be convincing me to join.

• 19:04Speaker 00

  Was that your first time applying your love of like numbers and engineering and, and programming to finance?

• 19:12Speaker 01

  Well, I was doing trading from my dorm room. Like, I was trying to be like someone like Ken or Jim Simons, you know, when I was there. So, I mean, it didn't work out that well.

• 19:22Speaker 00

  Yeah, were you good at it?

• 19:24Speaker 01

  Like, like trading. As a college student, not particularly, I mean, I found some strategies that worked okay. And then Ken actually told me like, yeah, these strategies aren't as good as like, like some of these strategies used to work. You back test them, but they don't work anymore because I'm arbitraging them away.

• 19:40Speaker 00

  So you maybe, maybe you and Ken found these same strategies at the same time, but Ken already had like a whole.

• 19:45Speaker 01

  I found them like five, six years too late, but yes, because I, because I had some good back tests, but in, in real trading, it was like close to zero.

• 19:52Speaker 00

  Okay. So you did find something, but Ken was like, neener, neener, neener. I got there first.

• 19:56Speaker 01

  Yes, yes.

• 19:57Speaker 00

  Is that, is that why he hired you?

• 19:59Speaker 01

  And that's fine. I mean, the he like, I wasn't trying to, I was just like trying to find interesting stuff to trade, right?

• 20:04Speaker 00

  Right, right, right. Is that why he hired you?

• 20:07Speaker 01

  Well, I think that, you know, he, part of like the, his rationale for why I should join is that like, it's much better to find these sources of alpha on a team. Versus doing it on your own, which I think is correct.

• 20:20Speaker 00

  Why is that correct?

• 20:21Speaker 01

  You know, when you're finding alpha, you have to bring a lot of different ideas together. And it's like unlikely that one person is going to see all of that at once. And we're, if you have like. A team of people with somewhat different backgrounds and approaches like, because like, like for, like the markets are very close to being efficient. And so like to find inefficiencies, like a bunch of different things all have to be true. And it's like easier to find those combinations in a group.

• 20:48Speaker 00

  Right, sure. Yeah. Add all the different perspectives together, kind of with the same lesson from Harvard.

• 20:53Speaker 01

  Just like, for example, if you were to tell me that, um, like you have a really interesting thesis on, you know, Ethereum. And then, you know, I don't know if I talked to Tom Lean, he gives me his thesis, then I put that together.

• 21:04Speaker 00

  Now you have two theses. Yeah, right, right. How long did you work at Citadel?

• 21:09Speaker 01

  Yeah. So I was, I was there for around a year and then another firm recruited me to kind of come in and. Continue, you know, to kind of actually, actually build a trading desk there. So I kind of, because I want to ultimately like, I wanted to. Do something more entrepreneurial, like I was trying to do that in college.

• 21:26Speaker 00

  Did you know that you wanted to be an entrepreneur?

• 21:27Speaker 01

  Yes.

• 21:28Speaker 00

  Yeah?

• 21:28Speaker 01

  Yes.

• 21:29Speaker 00

  When did you first know that?

• 21:31Speaker 01

  When I was eight.

• 21:32Speaker 00

  Oh, really?

• 21:33Speaker 01

  Yeah.

• 21:34Speaker 00

  What, what happened when you were eight?

• 21:35Speaker 01

  My dad and I watched a TV program about Bill Gates. And, you know, that was, I thought that was pretty cool that you could like, you know. Build software and, you know, create a great business around that.

• 21:47Speaker 00

  So building a company has always been something that you wanted to do.

• 21:51Speaker 01

  Yes.

• 21:51Speaker 00

  Yeah. Okay. And then you were kind of like searching around throughout Harvard and working with Ken. And then the next firm.

• 21:56Speaker 01

  I don't know. I mean, right. I mean, it's not like, that was the goal ultimately. That doesn't mean I had to do it right away. And then it's like joining a firm and learning a lot of the lay of the land of an industry is important. I mean, I guess, like, if we're talking about. Interactions with like famous people, one other interesting thing was like, when I moved from finance to Silicon Valley, like. I met with Peter Thiel at that point and his advice was like, don't start a company yet, like get to know how startups work first.

• 22:26Speaker 00

  Right.

• 22:26Speaker 01

  That was great advice too.

• 22:27Speaker 00

  Right. Yeah. And I'm assuming you were, you as, as an entrepreneur, you were always kind of like looking around for an idea. But you don't want to force an idea. You want the idea to come to you, correct?

• 22:38Speaker 01

  That's right. That's right. Well, I mean, you're looking around at some level, but I mean, when you're like on a trading desk. And the markets, especially like we lived through 2008, 2010, flash crash, all of that stuff like. When you're in the thick of it there, you're not thinking about like, what's my startup idea four years from now going to be. And now some of the information that you're seeing in the markets may end up. Like, you know, your brain may, may, may process that later, like in the showers, like, oh, actually like this is pretty cool. You can like. Maybe there should be like a better risk management platform or something, but that may happen years later.

• 23:11Speaker 00

  Yeah. How did you meet Peter Thiel?

• 23:13Speaker 01

  Uh, well, Peter, I met through, you know, through my friend who was CTO of Facebook, right? So.

• 23:18Speaker 00

  Ah, okay.

• 23:18Speaker 01

  Peter was their first.

• 23:19Speaker 00

  You're just connected there? Yeah. Do you have an ongoing relationship with him or?

• 23:22Speaker 01

  They, you know, the last venture round we, we did before launching a token, you know, they co-led that one.

• 23:28Speaker 00

  That was founders fund?

• 23:28Speaker 01

  Yes.

• 23:29Speaker 00

  Okay, okay. Okay. So then worked at Citadel, got picked up by a new firm to build out the trading desk. So took what you learned at Citadel, applied it to the.

• 23:38Speaker 01

  And it was a little bit like they, you know, nowadays.

• 23:42Speaker 01

  You know, big companies use the terminology like startup within a startup or something, or startup within a big company. And like. For me, building a trading desk within a larger firm was literally that, right?

• 23:50Speaker 00

  That was your first experience actually building a squad and being a leader.

• 23:55Speaker 01

  That's right. And, and just like building a business in the sense that like. Yeah. I mean, we, we were like, you know, I didn't have to set up HR or payroll or like.

• 24:03Speaker 00

  The silly stuff. Yeah. Right.

• 24:05Speaker 01

  So, well, yeah, I mean, it's actually, all of this stuff has some interesting aspects. So, but yes, like the, or like, I didn't, like, we had obviously relationships with. Like prime brokers, but I remember like on my first day doing that, like calling like data vendors, okay, like who has data for, you know, order like. L2 order book data for like this market. Like I was like doing that, right? It's not like, it was all there.

• 24:25Speaker 00

  What firm was this that you were?

• 24:26Speaker 01

  Graham Capital.

• 24:27Speaker 00

  Graham Capital?

• 24:28Speaker 01

  Yeah. Okay. So not far from where we are right now in New York.

• 24:30Speaker 00

  Oh, yeah. Yeah. Yeah. And that's also like high frequency trading stuff.

• 24:33Speaker 01

  Well, that's what they wanted me to build. They didn't have it before.

• 24:37Speaker 00

  I see. And so you built that?

• 24:38Speaker 01

  Yes.

• 24:39Speaker 00

  Yeah. When did you end up at Adipar?

• 24:41Speaker 01

  So, right. So then 2012 is when I kind of took the swing to go from. You know, Wall Street to Silicon Valley. So initially I was at Quora for a year and a half because. It was really interesting in the problem space there of kind of like, I guess I'm attracted to like matching problems in general. Like in the core, we're like matching users with questions and answers that to show them. Adipar was a really.

• 25:08Speaker 01

  You know, at the time, like fast growing fintech company, right? Kind of building actually like. Aggregating data from various sources to build ways to analyze portfolios like. And like a lot of, Adipar was started by Joel Landisdel, right? Who co-founded Palantir. And he's like, actually like. Applying some of these ideas to finance. Like if you look at 2008, a lot of the reason why risk was miscalculated is because a lot of the data that should have been there in one place wasn't.

• 25:31Speaker 00

  And so the, you were head of machine learning at Quora.

• 25:35Speaker 01

  That's correct.

• 25:35Speaker 00

  Yeah. And then also at Adipar.

• 25:37Speaker 01

  Well, Adipar was head of all of engineering.

• 25:39Speaker 00

  Head of all of engineering. And so this kind of goes back to what you were talking about with like the skills that you were uniquely good at. In your Olympia days, when your math competition days, where it's like, you really liked getting your hands on data. And that's, that's where you were particularly good at. And then Quora has like, here, we have all the, we have users looking for questions. We have people providing answers. How do we match these things?

• 26:00Speaker 01

  That's right.

• 26:00Speaker 00

  And that was, that was a big question to answer at Quora.

• 26:03Speaker 01

  Yeah. And you know, machine learning, so interesting enough, right? Like actually worked on. AI, what's now called AI, going back to 25 years ago. So, because I did like, in addition to the Olympias, also did a research competition called the Intel Science Talent Search. That was actually more prestigious competition than the Olympias. Like that's where you get to meet the president and, and stuff like that. And so the project I did for that was essentially like. Using optics to, instead of chips to like run neural networks faster. And but, so that, that's like the, you know, that whole field was like the precursor to, to deep neural networks, which then, you know, called deep learning. And then that lets like transformers and kind of modern AI.

• 26:46Speaker 00

  Was part of that actually physical? So using optics.

• 26:49Speaker 01

  Well, my project was physical, yes.

• 26:50Speaker 00

  Yeah. So part of the competition was a hybrid of being able to do math and coding. But then also build physical.

• 26:59Speaker 01

  You can do whatever you want. You can submit any research project you want. So my project. Like, I think probably at least half of the kids who competed in that did do some actual like. Hands-on stuff in the lab, but the others, you know, you could just write like. You know, like I placed 12th in that one, but. Like the guy who placed a second was like basically just wrote a complete theoretical math paper.

• 27:21Speaker 00

  Okay, I see. But what you did was one part physical with the optics.

• 27:25Speaker 01

  Yes.

• 27:26Speaker 00

  And so you had.

• 27:27Speaker 01

  Basically what's now called photonics. That wasn't a term back then, but it was just like optics and neural, neural networks. Yeah.

• 27:33Speaker 00

  Yeah, right. Okay. And it was just really trying to balance the actual optics, engineering and like all the math behind it to come up with a machine learning program?

• 27:41Speaker 01

  Well, it's to implement neural networks. Like, you know how, I mean, to be honest, like the project would be relevant today too, because like. Instead of throwing all this compute at silicon chips, like using photonics, like that is a thing. I mean, I mean, I didn't come up with that whole idea. I was like optimizing a particular aspect of it.

• 27:56Speaker 00

  Right, right.

• 27:57Speaker 01

  But the idea of using optics instead of silicon is a thing. And I think is like. That is a new form of computing, as is, you know, quantum.

• 28:04Speaker 00

  Right, right, right, right. And this, this was early stage machine learning when you were doing this?

• 28:08Speaker 01

  Well, not really. I mean, machine learning was around since the eighties.

• 28:10Speaker 00

  Sure. Yeah. Yeah. Well, I guess early stage by how we kind of know it today, which is just like, you know, AI.

• 28:15Speaker 01

  Yes. But what's interesting is like that competition. Back then, I think now these Olympias have become more famous because of, you know, success of some people have done them. But like back then, this research competition was considered like the main. Science competition in, in the US. And we got to meet the president, the vice president, a bunch of Nobel laureates. And the Nobel laureates, you know, and they were, I think, expressing the opinion of the scientific community at the time. They were like, yeah, all this like machine learning stuff, like that's a dead end. Like you, like that's kind of, it's cute that you did this work as a kid, but like. If you want to do science, like do quantum, like this neural network stuff is not going to work.

• 28:53Speaker 00

  Wow.

• 28:54Speaker 01

  And so, yeah.

• 28:55Speaker 00

  Did you, did you have a reaction to that when they said that?

• 28:58Speaker 01

  Yeah. I mean, I, at the time, I mean, I was like, who am I? I was like, yeah, like, let's, let's, let's focus more on the quantum stuff. Sure. Like, you know, I was, I kind of, I didn't move away from it. And then I came back to it 12 years after, when kind of in Marola at Quora.

• 29:10Speaker 00

  Yeah.

• 29:12Speaker 01

  I guess the, like the big data kind of changed the game. Cause I think they, I mean, they weren't wrong with, they were saying, given what exists at the time, because this was pre-internet scale of data.

• 29:20Speaker 00

  Right. Yeah. They didn't have the, you'd have the clay to do the machine learning to actually manipulate and stuff into being useful.

• 29:26Speaker 01

  Yes.

• 29:27Speaker 00

  Okay. And so tell me, I'm assuming, I'll make an assumption here, but I'm assuming there was like an aha moment or like a neural, neural connection. Coming together in your time at Quora, where you realized that, you know, you actually, I do have big data here. And you're doing the, the matching, the matching problem, like connecting people at Quora. And now you have the data and you have your experience with machine learning.

• 29:48Speaker 01

  Yes.

• 29:49Speaker 00

  Did some, something collide there? Did particles collide there in your head when you realized that, oh, there's actually a lot of. Clay here to work with from the machine learning perspective?

• 29:57Speaker 01

  Machine learning, you know, we were using it in quant finance too, because after 2008. A lot like before that, you could trade in fairly simple ways and it was more around implementing like fast execution. You know, what's now known as high frequency trading, right? But like pretty simple strategies worked. Then after 2008, those simple strategies stopped working. And you had to find more complicated patterns in the market. So machine learning was used for that too. It just wasn't. See the differences like in, in the tech industry, if you come up with anything interesting, you want to like create buzz around it and all that. In quant training, it's the opposite, right? Like you don't like, if you actually have really interesting math model to trade.

• 30:32Speaker 00

  Yeah, you keep it secret.

• 30:33Speaker 01

  And you want the competitors to think that you're doing something very basic, even if you're not, right?

• 30:38Speaker 00

  Right.

• 30:38Speaker 01

  So like, so we were using a lot of this machine learning stuff. Whether it's my team or other teams at my firm or competitors, like we're using a lot of this stuff. So I knew some of that already, including from like the work I did as a kid. But like, yeah, with with with Quora, you know, the internet companies, we certainly weren't the only ones to do it. I mean, Spotify was matching users and music recommendations. The guy worked on that is another kid I knew from Olympiads. Now he's running a company called Modal. You know, Netflix, you know, this, you know, if you recall, there was a Netflix Prize back in 2010. That was when there was like a million dollar prize to cope with a better algorithm to rank movies.

• 31:18Speaker 01

  So this kind of stuff like, you know, structured learning of like user data and user preferences and products to recommend. Like that was the thing. I mean, we were the first to do it in the space of knowledge at Quora. But what was funny is that Quora actually had a lot of the things you need to build L L M's, right? Because it was text data, unlike Spotify or Netflix.

• 31:39Speaker 00

  It was text data and it was also super information dense.

• 31:42Speaker 01

  Yes.

• 31:43Speaker 00

  Because it was all questions and answers.

• 31:44Speaker 01

  That's right.

• 31:45Speaker 00

  Yeah.

• 31:45Speaker 01

  Like in principle, we could have built the first good L L M. I mean, like our founder, you know, joined the board of OpenAI for recently, right? But like, you know, we, I mean, at the time, language models were pretty bad, though. And so the, we tried using them even like 12 years ago, they didn't add much alpha to the prediction model.

• 32:02Speaker 00

  Yeah, yeah. So you also had the connections to the two Anthropic founders from the Olympiad days, correct?

• 32:08Speaker 01

  Yes.

• 32:09Speaker 00

  And I don't know if how, where, how you kept up with them or if you kept up with them throughout the years. But you had those connections early on. You worked with machine learning relatively early on. You had your time at Quora. Was it odd when when AI finally happened, like we had the chat GPT moment in 2023 or whenever that was?

• 32:26Speaker 01

  Yeah.

• 32:27Speaker 00

  For you, was it like, oh, finally, like, we've got this thing now? Or like, what was it like being on, having the kind of like the information that you had?

• 32:35Speaker 01

  Right.

• 32:35Speaker 00

  Pre chat GPT.

• 32:37Speaker 01

  These companies had like interesting histories, right? So like OpenAI was started and, you know, they had like a bunch of co-founders, right? Like Elon funded it, you know, it was like, I think they're working on a bunch of, it was like pretty open ended lab, right? Like they're working on a bunch of different things. I mean, there was DeepMind back then, there was Google Brain. Anthropic was actually like, its team was interesting because some of them were actually academics up until like three years before starting a company. So that's probably the fastest anyone's gone from like being an academic to like, you're building a great business. So I think I got a lot of the big picture, right? Like that AI was going to be big and investing in AI first companies was going to be a thing.

• 33:19Speaker 01

  And so like the exact path that it took, I don't think I would have predicted it or I don't know if anyone predicted. Like the fact that, because even within OpenAI, like the L L M stuff was not like, I remember they gave a presentation in 2018 to the startup. We were part of this group called South Park Commons and we were using AI for what we were building at the time, right? Which was Lunch Club, like matching, again, matching problem, like kind of matching people with other people for professional connection. And anyway, so there were all these like presentations people gave, either people who were part of our startup community or external guests, right? So like the OpenAI guys gave a presentation, it was actually Dario who did it.

• 34:02Speaker 01

  He was still at OpenAI back then, right? And so they were talking about the stuff that where they were competing with DeepMind on like playing video games. And like, I'm not a video game person. And so I thought, okay, this is interesting, but like, this is not exactly the application of AI that I'm excited about. The reality is that they, they were working on that kind of stuff. They had like two guys on their team who were doing this L L M stuff. Like even with an open app, that wasn't their main thing.

• 34:27Speaker 00

  Right.

• 34:27Speaker 01

  And then they found some application that like was face shift different than like, I think G P for when they went from like GPT one to GPT two. It was pretty big, like, like if you're in the field, it all feels iterative, right? It's like the, the stuff gets better and better. But for the, for it to be interesting to the consumer was more of a step function. Because like, like, I guess when I first got a sense that this was going to be a big consumer product was in, was probably like two years. I had like, in late 2021, I remember some of our interns were like, you know, we don't really do homework anymore. I was like, what do you mean? Like this like GPT two is actually like good enough to do our homework. I was like, that's even just that is a big market, right? Just like homework.

• 35:12Speaker 00

  Doing homework, yeah.

• 35:13Speaker 01

  Yeah, I was like, I would, even, even if that's all it is, I was like, this is like, there's some product market fit there. Like this stuff is not just pure research anymore. But like, that's like when I first thought, okay, this stuff is going to work out. Because there's a lot of even within AI, right? Like even within what's used now, there's like generative video images, like all these things are quite different.

• 35:30Speaker 00

  Right, yeah, the whole kind of concept of AI has blossomed into basically everything.

• 35:34Speaker 01

  Yeah.

• 35:35Speaker 00

  I want to, I want to get to your time at Lunch Club. First, I want to talk about your time at Adipar.

• 35:39Speaker 01

  Yeah.

• 35:39Speaker 00

  Which is the, as I understand it, the last company you were at before, before Lunch Club. What was Adipar?

• 35:45Speaker 01

  So Adipar, as I was saying, was like the fintech company that Joe Lanzale co-founded.

• 35:49Speaker 00

  Right.

• 35:49Speaker 01

  That brought together a lot of financial data and like, you can understand your portfolio, like run analytics, run reports. What, what, what I wanted to also do there is to use AI to think about like what should be in your portfolio. At the time, the tech wasn't ready for that. So it was more around like understanding the, what it is now versus what it should be, but also very, very important function. Because if you don't know what's in your portfolio now, you think, well, why would you not know? Well, because if you have many layers of like, like maybe somebody owns Apple stock directly, but they also own an SPV. They may also own a mutual fund that has Apple, like under, like the correlations are actually not obvious.

• 36:27Speaker 00

  And you ran an internship program there, so you hired interns to work at Adipar, correct?

• 36:32Speaker 01

  Well, the, you know, it wasn't my idea, you know, Joe was hiring interns from day one. I mean, they had interns at Palantir too. So the idea to hire smart interns was not new. What was new that I brought to the table was that we should source them from the Olympiads.

• 36:45Speaker 00

  Okay, okay. So you, you picked your, an arena that you were familiar with to hire talent.

• 36:50Speaker 01

  We, you know, we were open to talent from anywhere, but like, we're specifically, you know, that was a like, you know, because we, it was like, it wasn't just me. We had a recruiting team and, and what I kind of was getting the recruiting team to learn about is like, hey, these Olympiads, you can actually like, you know, it's legal to work in the US, you know, 17 years old. Like we can hire people at any age who did, who did these Olympiads, it's going to be high signal hiring.

• 37:17Speaker 00

  Did that thesis work out for you?

• 37:20Speaker 01

  It did, it did work out.

• 37:21Speaker 00

  Yeah. How, how would you gauge that success?

• 37:23Speaker 01

  Well, certainly it worked out fairly well for Adipar, but it worked out really well for those kids.

• 37:28Speaker 00

  Yeah.

• 37:29Speaker 01

  And I mean, I think, you know, it worked out for those of us who Angel invested in their future projects.

• 37:34Speaker 00

  Yeah. Who, who were some of those kids that you, that you worked with?

• 37:36Speaker 01

  Well, one of them, I co-founded Lunch Club with, right? Scott, Scott Wu, who is now co-founder, CEO of Cognition. But yeah, you know, like between, between the Adipar intern program and, you know, the, the core intern program, I mean, like, I think, like, so, you know, co-founders of companies like Scale, Perplexity, Modal, as I mentioned. You know, that's just, that's just an AI, right? Also founders of some crypto projects too, you know, a bunch of them went to trading firms. I mean, I, I think it was, you know, very, I mean, I think, but essentially we just like, got a lot of the, the early stage talent in one place.

• 38:21Speaker 01

  And they, they just.

• 38:23Speaker 00

  Like shooting fish in a barrel.

• 38:24Speaker 01

  Yeah, they, they, yeah, and then like, they learn a lot from the full-time folks, but they also like form connections with each other. And some of them like started companies as a result of connections there.

• 38:36Speaker 00

  This is, we're going backwards a little bit, but to what degree do you ascribe the reasons why there was such a strong hit rate of talent coming out of the Olympiads?

• 38:47Speaker 01

  It was more of like, just like undiscovered alpha, right? You know, because I think from my perspective, it always would have been like that. I mean, maybe some ebbs and flows depending on specific cycles, but like, like for one, one piece of context, right? Like, so when I was at Citadel in the early 2000s or the mid 2000s, I guess I should say, like 2004, 2005, like, you know, and Citadel had a great recruiting team, one of, you know, top firms in the industry. But like, when I talked to the recruiting team, like they didn't know what the Olympiads were. Like that's crazy to think now, because now all these firms like, that's like all they do is like focus on this talent pool. But back then, like I told them from like, from first principles, like what these competitions even were.

• 39:32Speaker 00

  And so it's just you, you had the experience because you went through it and so you knew that it was.

• 39:35Speaker 01

  And I mean, I did it there. I mean, I'm sure there are other, you know, there was a guy at early Jane Street who was an Olympiad guy, like H R T, like all these training firms and startups too, right? Like, you know, CTO of Facebook did Olympiads, Google, their first CTO actually also did Olympiads. Like, so they, they, those people kind of slowly but surely educated their colleagues about this stuff, but it took years.

• 39:58Speaker 00

  And then the alpha got, got squeezed out.

• 40:00Speaker 01

  Yeah.

• 40:02Speaker 00

  Let's talk about Lunch Club. So Lunch Club was the first company that you founded, correct?

• 40:06Speaker 01

  That's correct.

• 40:06Speaker 00

  Yeah. Tell me about just how the spark came to you about like why Lunch Club was a good idea?

• 40:12Speaker 01

  Well, I think a lot of the, if you, like even if you just, if you think about our conversation right now, I think it's, it's pretty clear that connections are very important, right? Sometimes it's just like, even a single connection can change the trajectory of someone's life or career. In some cases, an industry, right? Like, you know, like what, how much value was created from Peter Thiel meeting Elon Musk in 97? The industries we know, this happens all the time, right?

• 40:36Speaker 00

  Right.

• 40:36Speaker 01

  So the idea was like, okay, but this stuff is all random. People just happen to meet. Like, there's all this data on the internet that we use as the data to shop. We use it to listen to music, to find recommendations for movies, for content to read. Why not use it as a way to like, do really valuable networking, right? Making the right connections and, but the, the, so that, that idea was kind of, that was the, the mission, but the implementation detail that made it work, or at least made it work as well as it did. Cause there, there have been others that have tried this kind of stuff, right? That didn't really go anywhere, but, I mean, for us, it worked pretty well, especially during COVID. And even since then, it's like plateaued and it didn't die, but

• 41:17Speaker 01

  the key insight on the implementation side was like, how do we avoid adverse selection, right? Cause if you have a system where you use like the, the model from, you know, dating sites where you're like, okay, like, like double opt in, that doesn't really work too well for networking because unlike dating where there's like somebody for everybody, like in networking, it's like, if you see, if you could send requests to anybody, like, it's like, you know, I don't know, Mark Andreessen would get like a million requests.

• 41:41Speaker 00

  Right.

• 41:41Speaker 01

  And someone who's like a college kid might not get any.

• 41:44Speaker 00

  Right.

• 41:45Speaker 01

  Even though what should happen is the college kid who's thinking about biology maybe should be matched with a college kid thinking about AI and they can make the next great AI agent for healthcare.

• 41:55Speaker 00

  Sure.

• 41:55Speaker 01

  But like, that's the kind of connection that should happen. Instead of both of those kids being matched with Mark Andreessen and Mark never having the time to respond.

• 42:01Speaker 00

  Right. Yeah. Right, right, right. Yeah. Cut out, cut out the intermediary or the indifferent and irrelevant point of the data.

• 42:09Speaker 01

  To be fair, Mark actually is very good at connecting people. So I don't want to use, you know, maybe a bad example of, you know, but, but I think you get the point, right? So adverse selection, so you actually don't opt into a particular connection. You let the system make the match.

• 42:21Speaker 00

  Right, right. So this was your first company and it seems to me just like a pretty logical continuation. Cause first you were working at Quora where you were matching between queries and answers.

• 42:30Speaker 01

  Yes.

• 42:31Speaker 00

  People and their queries and their answers. And then, and then you were working at Adipar, working on like, you were working on your own thing, but also the-

• 42:39Speaker 01

  Aggregating data than matching data, but, but yes. And then I guess a lot of the, like, I guess less on the product side, more on the people side of Adipar of like building these teams. I guess a lot of that was kind of irrelevant. Yeah.

• 42:50Speaker 00

  Yeah. And, and then, so then came Lunch Club, which is just kind of just the, you smash those two things together.

• 42:56Speaker 01

  You know, I should point out like, a, a large part of that was also deference of Scott too, right? So it wasn't, you know, it was only-

• 43:05Speaker 01

  All come from me by any means.

• 43:06Speaker 00

  Yeah. Yeah. Yeah. But it just seems kind of, kind of like we're kind of, your arc, the Vlad of the arcs, the arc of Vlad is kind of crescendoing here. At this time, it's, it's feel like you're going, you're kind of shifting from, you know, acquiring skills and competing in skills to, to applying skills. That's kind of like the inflection point that I see with Lunch Club. It was like, okay, Vlad, Vlad is like, kind of starting to find where he wants to really be an entrepreneur.

• 43:32Speaker 01

  Yes.

• 43:32Speaker 00

  Yeah. Talk about the transition from Lunch Club to Lighter, because this is the same company, correct?

• 43:38Speaker 01

  Yeah. So in terms of, you know, the, the process that it took, like corporate entity wise, it is the same. And I mean, I can talk about that as well, but I think the kind of like why lighter, I think maybe is maybe even more interesting question. Cause we, even like in 2017, when we settled on Lunch Club as idea, like we were thinking about crypto ideas then too. Like, I think when we had a shortlist of like five ideas, like two weren't crypto. So, you know, it's not like, if, if you just look at Lunch Club, like why would you go from this to crypto? But if you were to see the full picture of our IDMA's, crypto was always in the background. We just, like in 2017, a lot of the tech just wasn't ready to do what you need to do. Just how like 2014 AI wasn't ready for, you know, Adipar.

• 44:23Speaker 01

  But like, anyway, and I was really liked, like you asked earlier about what kind of math problems I really liked. And Olympia was like, one of the fields I really liked was number theory, right? Cause number theory had a lot of these like opportunities to do stuff really creatively and out of the box thinking. And so like, when I read the Bitcoin paper, 2012, again, the only reason I knew about it is because of a friend I made through Olympiad's, showed me the paper. So I was like, this is actually a really cool way to use number theory and your cryptography. So anyway, so kind of, I was always intrigued about the space and especially as it can address some of what I saw as missing in finance and being on TradFi, right? But yeah, the, the, the actual process we went through though, cause this is, it all sounds good and well now. It's like, oh yeah, we, you know, we did this and we did that.

• 45:09Speaker 01

  You know, it was, it was not easy, right? And like not a lot of startups, like, we didn't think it was like a sure thing that we could succeed with a pivot. But our, you know, like, I think in, in, in the space where now there's a lot of like negative, negative around VCs.

• 45:26Speaker 00

  Yeah.

• 45:26Speaker 01

  I mean, we wouldn't have been able to do that without the VC supporting the pivot. Because if you think, you know, this was in 2022 and the markets were weak across pretty much every industry. It was right after the collapse of kind of earlier crypto, even before AI started to pick up. So, you know, it was not a good environment for capital markets. Even companies like Robinhood were down a lot, right? Even companies like SaaS companies are from, you know, obviously crypto companies like Coinbase, everything was down a lot, right? So it wouldn't, wouldn't have been easy to just start from scratch. So we were able to retain 80% of our engineering team through the pivot. The way we did it was like, okay, like, let's have everyone be part of the process.

• 46:08Speaker 01

  Like, it's not just like, okay, well, we're, we're doing this. Now we're doing that. It's more like, let's run kind of like an internal YC within our company and let best idea win. So we actually were like testing three different ideas. Lighter was the one that was the best and we can, we kept building that and kind of took some interesting parts of the other ones merged that into lighter.

• 46:31Speaker 00

  Who came up with the lighter idea? Or how did that idea come about?

• 46:35Speaker 01

  I think that this one, you know, was idea that, that I came up with, yes.

• 46:43Speaker 00

  Okay. So the, what, what did you see in crypto that, that you, that really scratched your itch? Cause crypto is, I think from somebody who's very into numbers and into systems, probably provides a lot of material for you to, to think about. So what about crypto before you even committed to crypto?

• 47:00Speaker 01

  Right.

• 47:00Speaker 00

  What did you see that was like just intriguing or interesting to you?

• 47:03Speaker 01

  Right. So crypto, you know, like, I think there's a thing people talk about in startups. It's like a solution looking for a problem. And I, I think crypto kind of had a little bit of that. I mean, I guess Bitcoin was always providing value, but a lot of other stuff had more of the feel of like, okay, there's this really interesting tech, like there has to be applications for it, but like, you know, like, I mean, you were around 2017, 2018, a lot of the projects back then were like, you know, Uber on chain or like-

• 47:32Speaker 00

  Very squamorphic, very backwards. Yeah.

• 47:34Speaker 01

  Yeah. There's stuff like, okay, can we build like a Quora on chain or like, you know, can we like, build Airbnb on chain? It was all this kind of stuff, right? And it's like, there, there has to be, like this, this tech clearly works and it works to store value. Bitcoin, like there have, you know, it works as like, you know, decentralized computer with Ethereum, but like, there, there has to be like other applications of this that are really valuable, but it's like, hard to really put your finger on it. And so then the thing was like, well, like, let's look at exchanges. Like, why is it that we have these digital assets and they're, they represent decentralized protocols? For the most part, but the way they are traded, this was true when we started building, the way they're traded generally is not actually using the rails that they themselves are for.

• 48:19Speaker 01

  So they're, they're actually, yeah, I mean, there was Uniswap and, I mean, AMMs have their own inefficiencies, but, but that's certainly decentralized. But for the most part, like, that was less than 1% of the volume of how digital assets were traded. They were actually traded in ways that didn't actually use the rails that they were for. And so those, okay, like, how do we solve that problem? And we, we knew a lot about the tech already, right? But like, but the actual, like connecting that with, you know, cause you have to meet the customers where they are, right? It's like, you could build, you could sit in an ivory tower and say like, okay, how would you build the perfect system to replace TradFi?

• 49:00Speaker 00

  Yeah.

• 49:00Speaker 01

  Like that, that's not really how companies get built, you know? Unless, you know, unless you're Elon or something, but like generally like you have to start with a small customer base, build something that actually works and grow from there. And so like, that's where, okay, like perps is actually something people want. They want decentralized perps, but also that are secure and verifiable. Like no one's been able to do that. Like we, we think we can.

• 49:24Speaker 00

  Lunch Club pivoted into lighter in, in 2022-ish. The first time I ever heard about lighter was sometime late in 2024. So what happened in those two years?

• 49:34Speaker 01

  Well, we were, I mean, the core tech, the solution we came up with to the problem of building an exchange that's low latency, low cost, secure, verifiable and composable, right? Like that took 18 months to build that. Cause it had never been done before. We had to use ZK in kind of novel ways and come up with.

• 49:51Speaker 00

  So you guys were just heads down engineering, doing some like hardcore, cause nobody had ever built anything like that before. And so this was all just hardcore engineering work that you guys had done for like 18 months.

• 50:00Speaker 01

  Yes.

• 50:01Speaker 00

  Okay.

• 50:01Speaker 01

  Now, we were talking to some small groups of customers and like showing them prototypes or like different versions. But like, we certainly weren't publicizing anything that we're working on in, in, in the open. It was more like, I mean, we would go to conferences and meet traders and, you know, other builders and show them some stuff and build those relationships over time, which actually proved really useful, right? Because when we got to, when we did have the tech ready, we had a hundred traders ready to try it and they were, they weren't just like people who spent five minutes on it and left, like they, they knew us for a long time and they really-

• 50:37Speaker 00

  They knew the grit.

• 50:38Speaker 01

  Yeah. And they really like committed to testing out and gave it, giving us feedback over months.

• 50:44Speaker 00

  The lighter, the problem of lighter, as in the, the problem that lighter is trying to solve, seems to be quite the amalgamation of all the other things that you've been doing. Starting with just like the, the physics and the informatics and the math of Olympias. But then also just the matching engine work that you've been doing. You did at Quora, the high frequency trading work that you did at Citadel. It seems to be like the perfect problem for you. And maybe that's why you picked the idea in the first place, but-

• 51:11Speaker 01

  Yeah.

• 51:11Speaker 00

  And for a guy who's hyper competitive and just really into, into numbers and math, it seems to be the perfect substrate to solve a very fun problem. Is that how it feel?

• 51:22Speaker 01

  Yeah. I mean, it, it, it definitely feels great to have- You know, it's one thing to like work on something that

• 51:30Speaker 01

  that, you know, you feel like you're like, you're built to do it. I mean, it's something to actually build it and have real customers. All right. I mean, these things, there's like levels to this stuff as far as the satisfaction you feel. I mean, I think we're still very early. If you zoom out, so hopefully there'll be like more rewarding feelings in the future. But yeah, no, it's been great. I mean, I think the way we thought about the idea maze was like, want to build something that sits at the intersection of three things. One, like something that there's like a large market for, right? Another thing is a mission we're excited about. And third, something we would actually be, we would be like good at building, like we would do well against competitors if we built it. And so like,

• 52:15Speaker 01

  yeah, I mean, I mean, I think like Lunch Cup fit across all three at the time, although I think where we got it wrong was the market was actually not very big. Like it got big during COVID and then shrunk.

• 52:25Speaker 00

  Sure.

• 52:26Speaker 01

  But with with lighter, I think all three things are true. And it's clearly like trading. Drives is a huge market, you know, the mission of doing that in a decentralized and secure way is a mission we care a lot about. Or, you know, especially after I saw how TradFi works, what works, what doesn't work. And then, yeah, I think we're kind of particularly, you know, we have a particular set of skills that are good at solving this problem, specifically because we have people on the team that know a lot about cryptography and people who know a lot of quant trading. And like, if you're building a system for traders, including quant traders, which is most of the market makers are like, it's really helpful to have that that background.

• 53:05Speaker 02

  Some exciting news. We are launching a new podcast to help people figure out the crypto cycle, how to navigate it. The best crypto cycle investor I know, his name is Michael NATO. He runs the DeFi Report. This is the guy that sent me a sell alert before the 1010 price drop happened. His cycle analysis has been absolutely on point. I've been following him for years. And this year, we started recording weekly podcast episodes. Each one, we get into his portfolio, what he's holding, the market structure, entry targets, fair market value of Bitcoin and Ether, and where we are in the cycle. There's new episodes that are released every Wednesday. They're 30 minutes, they're short, they're punchy. I think this crypto cycle is harder to navigate than most, so let's do it together. Go subscribe to this podcast, search the DeFi Report, wherever you get your podcasts, YouTube, Apple, Spotify,

• 53:49Speaker 02

  or find a link in the show notes. There's a new episode waiting for you now.

• 53:52Speaker 03

  Hey, Bankless Nation, it's David. If you're hearing this, that's because you are listening to the free Bankless Podcast feed. Did you know that there is a premium Bankless RSS feed? The premium feed has extra interviews that I do for my own personal research and just deeper questions that I want answered about the crypto industry, questions that I want to answer so I can be more informed as an investor, both at Bankless Ventures and also just in my own personal portfolio too. Also, there are no ads, which means if you listen to the premium feed instead of the free feed, you'll get about 20 hours of your life back every year because you choose to support Bankless directly. So if you're interested in getting extra content all while skipping the ads, or you just appreciate what we do here and want us to keep doing it, we'd appreciate it if you signed up for Bankless Premium. And there is a link in the show notes to get started. Cheers to a good 2026.

• 54:34Speaker 00

  How would you articulate the mission of lighter? In a long form way, like what's the, what is the big mission of lighter?

• 54:42Speaker 01

  Coming up with like a short concise statement is like, you know, the Mark Twain quote is like, I wish, you know, I was writing you a short letter. I didn't have time to write you a long one. So that, I mean, that's, that's hard. I mean, I don't know if we've like cracked that. I mean, the way I like to think about it is these like five pillars, right? Like low cost, low latency, secure, verifiable, composable. And we can zoom in on which, what those are and why they're important. I mean, I think low cost and latency are pretty self-explanatory. I mean, I think secure, this is like after, you know, security isn't just theoretical concept. Like, we've seen, you know, we see different, you know, both in TradFi and and in crypto, right?

• 55:26Speaker 01

  I mean, we see like hacks and kind of, you know, even things like assets being seized, you know, there's like security and being able to like actually like hold on to your assets no matter what is really important, right? But verifiability is equally important, right? Because with verifiability, you wouldn't have Madoff, you wouldn't have, you know, FTX. You wouldn't have a lot of these things that have been a problem for for finance. And composable, that's, that's the one where like all these DeFi protocols, I mean, have been composed with, that's one of the nice things about DeFi, but I mean, that's not new, but what is new is like DeFi and TradFi being composable. Right? Like all these like financial primitives.

• 56:10Speaker 01

  You know, now we're talking about like tokenized stocks and perps and options all on the same balance sheet, like like composability, not just from like a software level, but from like a, you know, a balance sheet level is really cool. So anyway, I mean, I think this is all what does, how does this take shape over the years? I mean, I'm a little bit, I like what, you know, Jensen has to say, you know, from Nvidia about how like, you know, we don't think about five year plans, we think about what we're going to do tomorrow. So like, it's, I mean, it's helpful to think about like, what does this look like two, three, four, five years from now, but really like, I think we know one, two quarters out and we listen to customers and we keep building. But I mean, ultimately, it's like whatever the merge of DeFi and TradFi looks like, we want to build the tech to make it happen.

• 56:54Speaker 00

  Well, my next question for you is going to be, what does lighter want to be when it grows up? But I guess, I guess you think in in one or two quarters. And so you can only tell me what you're going to, what lighter wants to be in two quarters.

• 57:05Speaker 01

  Yeah, I mean, I think the, the saying like, the longer term kind of being the technology layer for, you know, DeFi meeting TradFi is a way to describe the mission for sure.

• 57:18Speaker 01

  But again, I think when you pick like a short statement like that, that misses something because then

• 57:21Speaker 00

  Sure.

• 57:21Speaker 01

  anytime you do that, somebody can be like, oh, well, what about this thing though, that I care about? Is that part of the mission? No, it is.

• 57:28Speaker 00

  It's also that. Yeah. Yeah.

• 57:29Speaker 01

  And so, but yeah, a couple of quarters out, it's like, you know, one of the really interesting things that I really like about kind of, you know, the crypto space is that like, you know, there's this question that comes up when you start building a company and software, it's like, are you a product company or are you an infrastructure company? And I feel like with crypto, you can actually be both. And not, not, not just to check the bugs, but you're, you're, you're actually building something that by its nature can be both. And so from that perspective, like, yeah, we're building lighter core and lighter core is interoperable with other instances of lighter that were, that are used by partners like Telegram Wallet, Robinhood, et cetera. There's kind of many more of those to come.

• 58:12Speaker 01

  So that's, that's big. But then also we're building products, you know, directly like, like options, like, you know, stuff with AI agents, like there's a lot of exciting stuff.

• 58:21Speaker 00

  Well, Vlad, we're coming up on time, but let me take a moment to just kind of articulate about what about lighter excites me. And maybe I can get you to react to it. And really been all throughout my time in crypto, we've seen an evolution of exchange technology. And first we had like the, all the different proliferation of blockchains in the fork and fair launch phenomenon in 2013. And we had this new crop of exchanges. We had, you know, the the bitrexes and the bitmexes of the world. And we had this growth of centralized exchanges. So we like, we unlocked that part of the tech tree. Ethereum came along, we had Ether Delta, which put that kind of technology, but put it on chain. Terribly slow, terribly inefficient, but nonetheless, it worked. Uniswap came along, we had the AMM, got a little bit better.

• 59:05Speaker 00

  We had the L2s come along and there's been this just arc of exchange technology that has grown. And I see, and like you had the big exchanges like Coinbase and Kraken and Binance kind of really plow the way of like what actually like compliant and compatible centralized exchanges works with with like TradFi and regulation and all this. With lighter, it seems to be that there is somewhat of a step function change in terms of the technology under the hood with the ZK circuits that optimize for latency, but also providing the, the auditability. That is something that Coinbase can't offer, Binance can't offer, Kraken can offer, Uniswap can offer, but Uniswap can't offer

• 59:49Speaker 00

  the speed and the latency of the ZK circuits. And so it's like, if, if Coinbase was like next generation exchange, and then Uniswap was like a next generation exchange. Seems like lighter is like something like Gen three or the, the most latest generation of exchange technology, which to me kind of embodies a lot of the ideals that crypto has, the user verifiability. But it also has the product demands that traders want and must have. And it seems to be there's just like a synthesis of all of the things that we've ever learned in crypto, all kind of being applied into the same spot to create a, create a product that has a lot of these, that satisfy some of the most hardcore users, the traders, but also people like me who are more like idealistically driven.

• 1:00:34Speaker 00

  And so that's kind of like, I just think of it as like the next generation. I think like the NASDAQ or the CME or the, or the New York Stock Exchange would hopefully one day be built on that same kind of like ZK circuit substrate. Because it's just like, it's just better for everyone. I'm sure the regulators also want it too. Because it's better for them as well, the transparency of it all. And so that's kind of like why I've gotten excited about lighter. And I just want to kind of like react, you, you can react to that however you like.

• 1:01:01Speaker 01

  Well, we're, first of all, like we're, we're glad to see, you know, really passionate, you know, support from, from early community members. I think you articulated it as well as anybody. So that's like, that's really rewarding as an entrepreneur, right? To hear kind of early users really understand. In some cases, better than we do why what we're building is important. But, but I think, you know, as a technologist, like you talked about how, okay, like there have been aspects of this that have worked well in the past, but there's always a trade off and like, we found a way that where we can actually kind of achieve all these things. And that, that's what's really fun, right? As a technologist, it's like sometimes, like most of the time you're solving problems where like you're optimizing X at the expense of Y.

• 1:01:44Speaker 01

  And so, you know, you're, you're making calls on like, okay, is that still like net, net, is that the right thing to do or not? But sometimes you find these solutions where you can actually improve X and Y or X and Y and Z. And that's from, I think you, those come about kind of once in a generation in a sense that like, you know, like I think in AI, like the transformer was such thing, right? I think in, in crypto, the ZK proofs were such a thing, where like with this technology, you can actually like do better across many dimensions relative to the previous frontier. And so, so that's kind of on the technical side, what it is. And I think right now we're the biggest ZK project, like I think arguably like lighter is

• 1:02:28Speaker 01

  kind of what proves that there's, that ZK has real applications.

• 1:02:33Speaker 01

  And then I think the second part of what you described was like how TradFi perceives it. You know, TradFi is interesting, right? Because they they understand the tech nowadays. They they're testing it out in various forms. Ultimately, you have to remember like TradFi follows the money, right? So when they see product market fit, they want to make sure the stuff is legit and works in ways that are fair and verifiable and, you know, many cases, regulated. So that's all prerequisite, but like ultimately TradFi also needs to see that it works. So you can't just go into TradFi and say, okay, like, you know, I have a better mousetrap. Like, you should just, you know, you're running trillions of dollars

• 1:03:16Speaker 01

  of, you know, volumes, and you've done it for 150 years, but I have a better way. You switch to my way. Like, it doesn't work. Like, they're gonna do crawl

• 1:03:24Speaker 00

  They're gonna tell you to go go build the mousetrap and then maybe we'll use it.

• 1:03:28Speaker 01

  Yeah, exactly. And then those like, well, maybe it's like, you know, like, maybe the path would looks like where you first try it. Like, I think the example of, you know, ICE making a strategic investment in poly market is like a good example of this, right? It's like, like, prediction markets weren't something that ICE was doing before. It's like, for a new space, they're gonna try new technologies. Now, if they work, they may explore it further. And so that's kind of, I think, how TradFi looks at things. It's not like, okay, this, I mean, similar to AI, right? Like, when if, you know, a large bank, it's not like large banks in 2023 said, okay, this AI stuff is interesting. We're gonna rebuild everything with it right now. It's more like, let's try it out in different pockets. Like, we'll use it for this part of the business, for that part of the business. And eventually, it becomes the thing. Simple. Yeah, yeah.

• 1:04:14Speaker 00

  And from the margins.

• 1:04:15Speaker 01

  That's right. Yeah. So that's how we see it playing out. But, you know, a lot of the stuff has to do with like, particular constraints that these businesses have. You know, they all have shareholders, they have their own management structures, this and that. And we just have to be willing partners, you know, to engage and meet them where they are.

• 1:04:31Speaker 00

  Well, Vlad, we've got a particular set of skills. It's been fun watching you apply them. And I will continue to watch you apply them over the year. So thanks for coming on the show and meeting me in New York.