A CS Education That's Free Until You Get a Job - Austen Allred of Lambda School

by Y Combinator4/3/2019

Austen Allred is the CEO and cofounder of Lambda School.

Lambda School provides a CS education that’s free until you get a job. They were in the Summer 2017 batch of YC.

You can learn more about Lambda School at lambdaschool.com.

Austen is on Twitter at @austen.


00:35 – Encouraging people to do something that they’re scared to do

4:50 – Where did the insight for Lambda School come from?

6:00 – College vs developer schools

10:00 – Building a network

11:50 – Does Austen see value in a traditional liberal arts education?

14:30 – Steven Klaiber-Noble asks – As competitors begin to copy your model what front do you believe you’ll be competing on?

17:30 – Why did Austen choose to raise money?

19:40 – Fundraising falling through on Austen’s first startup

21:10 – Moving back to Utah and writing a book about growth

23:00 – Why Austen wrote a book

26:00 – “Starting a company is by definition saying, I think what I can do is worth more than what other people will pay me for.”

27:00 – Mispriced human capital

29:55 – Other opportunities for Lambda School

32:20 – Modeling risk

35:10 – David Kofoed Wind asks – When Lambda School is incentivized to take in people that will land high paying jobs, how do you think about the diversity of candidates? One would imagine that it quickly becomes a game of pattern matching the stereotypical SV people.

37:25 – Will Lambda School ever not be remote?

40:50 – Dave Dawson asks – You appear to be on the successful path now, was there a point early in Lambda School when you wanted to stop?

43:20 – Helping everyone become an autodidact

46:20 – Rethinking where to start on an online course

47:45 – Dave Dawson asks – What keeps you up at night at this point?

49:20 – Dayo Koleowo asks – “I have made remarks I do not agree with” – from Austen’s Twitter bio. What is that one remark you wish you didn’t have to disagree with?

53:00 – Choosing remote work as a core problem to solve in your company

55:30 – Analysts aren’t good at measuring product quality

57:10 – Teaching taste


Google Play


Craig Cannon [00:00:00] – Hey, how’s it going? This is Craig Cannon and you’re listening to Y Combinator’s Podcast. Today’s episode is with Austen Allred. Austen is the CEO and co-founder of Lambda School. Lambda School is a CS education that’s free until you get a job. They were in the summer 2017 batch of YC. You can learn more about Lambda School at lambdaschool.com and Austen is on Twitter @Austen. All right, here we go. Today we have Austen Allred. He is the CEO of Lambda School, which was in the summer of 2017 batch and Lambda School is a CS education that is free until you get a job. Austen, what I wanted to ask you about, you mentioned this on a few podcasts that I’ve listened to you on and basically, there’s this core idea in Lambda School that you have to be kind of competent enough to even apply. This is a great opportunity for many people but what I’m curious about is how do you encourage people to do something that’s probably good for them that they might just be scared to do or they make excuses about?

Austen Allred [00:01:03] – That’s the actually the problem we grapple with the most. Patrick Collison said something along the lines of, “If you think about it, the highest leverage activity any human can have is inspiring other humans to do what they’re capable of.” In many ways I see Lambda School as the realization of that. We’ve had a lot of students, in the past you would graduate and then whenever you felt ready, you would start applying for jobs. And we’d have students that would just wait and wait and wait, and we’d be like, “What, you know, why are they not working hard? What’s going on?” You talk to them and you realize it’s massive imposter syndrome. They’re just not, they’re not confident or comfortable with the idea that they’re now a software engineer. They’re just kind of procrastinating that. We now we have some mechanisms that get you to start applying earlier. The minute they start applying, they get interviews, they get hired, they get a six figure job and it’s like, “What? What were we waiting for?” Right, you’re totally capable.

Craig Cannon [00:02:03] – What about the people before Lambda School? A few of the questions that came out from Twitter were basically like, “Isn’t this just an opportunity for rich people?” Being able to leave your job and commit full time to a seven month program?

Austen Allred [00:02:17] – My first answer to that would be, we have part time programs as well, so you can do it evenings and weekends. We have experiments running where we pay people living stipends that we’ll be expanding greatly, we’ll be announcing something in the near future along those lines in concert with a few other tech leaders here in the valley. The number one problem we have to solve on the admission side is getting the people who are qualified to actually apply. You wouldn’t think of it as being a problem.

Craig Cannon [00:02:52] – It’s a YC problem too.

Austen Allred [00:02:55] – Yes. Some of the best companies, there’s like, “Yeah, I just don’t know if I can get into YC,” and that’s a difficulty of having a good reputation where a lot of people want to go there. There are people who assume that they won’t get in, which is not the case.

Craig Cannon [00:03:12] – Right and so your acceptance rate, maybe it’s changed, but I heard you say it’s 25% of the people who make it through that first initial test.

Austen Allred [00:03:21] – Of the people that make it through the pre-coursework it might even be higher now. I mean the pre-coursework is more difficult, but the vast majority of our selection mechanism is the pre-coursework, of how well you do on it, if you put in the time. Sometimes I wonder if we could just make the pre-coursework really long and have that be it. No interview, no nothing, just the work.

Craig Cannon [00:03:44] – That’s obviously your business hinges on you assessing that risk, right?

Austen Allred [00:03:49] – Correct.

Craig Cannon [00:03:50] – If you could make it sufficiently difficult, it would work.

Austen Allred [00:03:54] – In theory. You run the risk of false negatives or false positives I guess. All that we need to find out is are you of average mental intelligence and will you work really hard? YC is similar, right? When I got into YC, I felt super intimidated. There were, and there are brilliant people all over in the room. But just because you have a PhD in computer science from MIT and you got a 1600 on the SATs as a freshman doesn’t necessarily mean you’ll have a successful company. It’s not directly correlated.

Craig Cannon [00:04:33] – In the past it’s possible that YC was fooled more often by fancy degrees and PhDs and stuff like that.

Austen Allred [00:04:40] – Possibly, yeah.

Craig Cannon [00:04:41] – Not the case anymore.

Austen Allred [00:04:42] – I had in my YC group, there were a couple, there, like a 15 and a 16 year old.

Craig Cannon [00:04:47] – Oh, you were in that group? Yeah, that was a fringe case. I’m curious about… There’s this core insight with Lambda School, which centers around what if we didn’t have to charge 10000 dollars for this dev school? How did you reach that point?

Austen Allred [00:05:07] – I was living in rural Utah before I moved to San Francisco. When you move to San Francisco from a small town and you see just the discrepancy in opportunity I guess. It was easy to look around rural Utah and be like, “Look, that guy is just as capable and intelligent as that guy in San Francisco. But in Utah, that guy’s making 30K, in San Francisco, that guy’s making one 150 and he’s leading a team.” How does that work out? Is it just what jobs were available? If you talk to people, you realize it’s more about opportunity than anything. To tell a guy that’s making 30k, “Hey, just drop, you know, 10k on an education and there’s a 7% interest rate on the other side and hopefully it works out,” that’s actually a really big ask. I have to put it in terms for VCs sometimes and say like, “Imagine spending half of your annual income on an education on a bet.” Most VCs are very well off and you have that kind of risk that you can take. You have to kind of put it into that similar psychological context that people just can’t afford that risk.

Craig Cannon [00:06:19] – Why are our people in these situations signing up for University of Phoenix type programs but not the dev schools because it’s a similar choice, right?

Austen Allred [00:06:30] – The first reason is honestly because the University of Phoenix is more aggressive in their marketing tactics. They don’t really care whether you succeed to some degree. They’re backed by the loans are backed by Uncle Sam and whether or not you’re successful, they get paid. The incentives are fundamentally misaligned and if the University of Phoenix can get you to sign a piece of paper, they get 80 grand and that’s it. But the reason people opt for universities generally speaking versus code schools, is they’ve, you know, my entire life growing up, nobody ever said there’s a path to success that’s not college. And I’m not a college graduate, my older brother isn’t a college graduate, a lot of very successful people I know are not graduates, but yeah, everybody growing up just said, go to college, doesn’t matter what it costs, just get a degree, it doesn’t matter what you study and you’ll be successful. My generation is now realizing like, that’s fundamentally untrue. It isn’t only not good advice, it’s probably bad advice, net-net and the newer generation has the benefit of seeing people with hundreds of thousands of dollars in student loans and working at Starbucks. Clearly something was off in the advice that we were given.

Craig Cannon [00:08:00] – Wow, yeah yeah. I mean, I went to NYU, so I know the feeling. I graduated with like 50.

Austen Allred [00:08:06] – And that’s pretty good.

Craig Cannon [00:08:07] – I got a bunch of scholarship and it worked out all right.

Austen Allred [00:08:12] – But there’s still a lot of debt to be clear.

Craig Cannon [00:08:14] – Yeah, dude, I was an English major, so I got all the jobs I’ve gotten have been through skills I’ve learned outside of NYU. Obviously I can write fairly well,

Austen Allred [00:08:23] – For sure.

Craig Cannon [00:08:25] – But the real upside was being in a huge city where I could start working early and I think that was a real competitive advantage. I’m curious about Lambda School in this way because at NYU, once I started, I never went home again. Every summer I was working and then I was working during the year and then I was like realizing that I could cram all my classes on one or two days and work full time. And so I had easily a four-year advantage on all my friends who went to, you know, like Colgate or something like in the middle of nowhere, right. How do you take advantage of that at Lambda School? Getting people in early, getting them a network?

Austen Allred [00:09:01] – There are a couple of things. One is, Lambda School is nine months full time versus a four year degree, which is obviously four years. The average Lambda School grad if you had the two students, one started a university at the same time as another student started Lambda School, by the time the students graduating from university, the Lambda School student has three years of experience and has paid Lambda School off and has a quarter million dollars a head at least of the university grad. Net-net, it’s pretty obvious. The network is pretty key. Most of our students are coming in with no network, a lot of students from either inner cities or rural areas. We basically have to help them build that and then build a network around them. We have a full time team that is just bringing in companies to hire students, part of the school itself is actually, “Okay, here’s how you network, here’s how you find other people with similar backgrounds hired in engineering roles on LinkedIn. Now we’re going to go contact them.” That’s actually built into the curriculum.

Craig Cannon [00:10:08] – Oh, I didn’t realize that. Can you actually break those down? I’m very curious about how you tactically explain networking skills. I know you wrote a growth book before Lambda School, right. You’re into breaking these things down. Say I’m a graduate or I’m inland to school at what point do you start teaching me, I guess you could call it soft skills.

Austen Allred [00:10:29] – Now you meet your career coach at week three. It’s very integrated into… It is Lambda School, right, it’s not a separate external thing. And we have careers lessons every week for the entire school. For example, the, you know, we train students to do informational interviews, which is find someone with a similar educational background to you. Use LinkedIn’s advanced search and search for schools that are x, y, z in locations that are this and with careers that are this. Now you have a huge list of people who are just like you, reach out to them and ask if you can grab coffee just or jump on a phone call and say, “Hey, I want to break into this field. What do I do to break into this field?” Probably our best job offers actually come through that. Your best job offers always come through networking and meeting people. We’re really close to the point at which we just say if you have to use an application to like click an apply button somewhere, just throw it in the garbage. Don’t do that anymore.

Austen Allred [00:11:39] – Or at a minimum we’ll have somebody else do that for you. Now to help with that, we are hiring right now 37 interview sourcers. Those are folks that are just going out, they’ll take your specific information and they’ll go out and help find companies as well and bring the companies to you. We have an entire team that’s dedicated to that.

Craig Cannon [00:12:03] – Wow.

Austen Allred [00:12:04] – The future, we’re going to look a lot like a talent agency basically for software engineers and designers and whatever else.

Craig Cannon [00:12:10] – Right. In terms of the other, there are a couple of questions around, well, so John Palmer asks, “Can or what other pieces of the university experience can be unbundled?” The thing that I’m in particularly curious about is the way you treat education in the sense of Lambda School is like, “Hey, this is job training.” This is the goal of making money. It’s not about being altruistic and necessarily making artists or whatever you want to call it, right. Do you see a traditional, a value in a traditional liberal arts education and like those elements for school?

Austen Allred [00:12:47] – For sure. I see value absolutely in the liberal arts. I do disagree that the only way you can learn the liberal arts is within the walls of a university. A lot of the people that we’re talking to if you’re 35 and you’ve never had a very high paying career, the prospect of let’s take out 50000 dollars of loans to study Shakespeare is not a bad thing per se but probably not the right thing at the right time. We’re entirely vocational. We’re a trade school basically and we want to help you make as much money as you can and then when you’re making a lot of money, it’s a whole lot easier to make decisions around what you’ll learn and the liberal arts. In fact, that may be an add-on we add to the school later, but it’s just not the focus right now. There are a lot of people that do a really good job at that and we’re okay to leave that to them for now.

Craig Cannon [00:13:54] – It’s one of those things similarly to signing a student loan when you’re 18. You don’t really know the terms of the deal. You can read it on paper, but you don’t really understand the dynamic you’re about to enter. Whereas when you’re say, yeah, you’re 35 thinking about switching careers here, you’re much more practically minded. Or you might just be interested in indulging.

Austen Allred [00:14:14] – We’ve romanticized it a little bit. I went to the liberal arts portion of the liberal arts degree before I got into the specialization and like, it was valuable, but it’s, I’ve learned way more by reading books and meeting people and other stuff than I did in the university. I get the value, but I don’t buy the argument that like, that’s the only time and only place that you can become well rounded. That doesn’t make sense to me.

Craig Cannon [00:14:42] – Do you see there are valid, there are elements of value and unbundling other aspects and like really building that into Lambda School?

Austen Allred [00:14:50] – For sure.

Craig Cannon [00:14:51] – There’s another question here from a Steve Clarborne Noble who asks, “As competitors begin to copy your model,” which is we should talk about, it’s an income sharing agreement, “what front do you believe you’ll be competing on?”

Austen Allred [00:15:10] – All of them.

Craig Cannon [00:15:12] – All right.

Austen Allred [00:15:13] – The way I describe Lambda School is it’s… People know Lambda School because of the business model, but the business model is something that’s up front. That’s not what makes Lambda School work. You can charge whatever you want or however you want, the key is making that successful. Our product is, our educational experience is I think among the best in the world and it has to be the way we’ve built it in order to make a free upfront and online education work. When we started our dropout rate is what was killing us. Now we’re graduating, you know, more than 85% of students that start day one. A lot of the dropout is because of reasons outside of our control, it’s financial or life circumstance or whatever else. We have a product that people love, they come out on the other side as great software engineers and we have a hiring network ready to hire them. That’s all really difficult. Then the admission side of knowing who and how and what to accept. That’s not something you can buy. You just have to figure that out over time and eventually it will be very data driven. We’ll be able to tell on before a student begins Lambda School what they’re likely to be hired at and what they’re likely outcome will be similar to a lending company.

Austen Allred [00:16:39] – But there’s no credit score for people. You can’t say like, “What is this person’s potential?” There’s no inputs that you can take right now that can determine that. Nobody knows how to underwrite student risk. Nobody knows how to train students online. Nobody knows how to play students nationally at scale and figuring all that stuff out is the hard part.

Craig Cannon [00:16:58] – Right.

Austen Allred [00:16:58] – Plugging, throwing an ISA on the front end is the easy part.

Craig Cannon [00:17:01] – You’re just hoping that like you’re much further ahead than people realize.

Austen Allred [00:17:06] – Way further. I’m sure there will be competitors.

Craig Cannon [00:17:10] – Sure.

Austen Allred [00:17:11] – But the thing that scares me isn’t somebody adopting our business model that probably has happened that I haven’t heard about and will happen a million more times. What will make us win is the student experience and how fast we can move in making that better over time. That includes the outcomes that students have. All of my time and effort and energy is focused on that and that’s really freaking hard.

Craig Cannon [00:17:39] – Okay. It’s way hard. You were profitable before you did YC, right?

Austen Allred [00:17:47] – Correct.

Craig Cannon [00:17:48] – Alright, yeah. Maybe not usually profitable, but profitable.

Austen Allred [00:17:51] – We are profitable by any definition.

Craig Cannon [00:17:54] – Yeah, there you go, okay, cool. Was there an element of fear around being copied that led you in the path of raising money?

Austen Allred [00:18:05] – No. We actually sat down with a YC partner and we said, “Okay, we don’t need to raise money right now but if we do, we’ll suffer x percent,” so we modeled it out. We actually created a financial model and said, here’s the dilution that will suffer and with that money that we’ll raise, we’ll be able to grow like this. Therefore, you know, and it came out to something like Lambda School will be 10 times bigger than it otherwise would be, and it will suffer, call it 15% dilution. That’s an easy trade off if you actually model it out and that’s the reason we have continued to raise. Before we started Lambda School, I was like, I’m never raising VC again.

Craig Cannon [00:18:53] – Right, well we should tell us straight up that was a cool story.

Austen Allred [00:18:56] – Yeah and now, it’s a year and a half later and we raised 48 million and it’s not going to be the last we’ve raised. If you look at the growth rate of the school and the revenue, if you look at any number, all of those raises were wise decision in retrospect. There’s a reason VC exists. When you have a company that can grow at extreme scale and capture a market and create way more value that’s why you raise VC. The realization for us was realizing that Lambda School could be a successful bootstrapped company, but if we raised and use a little bit of fuel, we could just be much bigger and much more impactful.

Craig Cannon [00:19:43] – I actually don’t think it’s dissimilar from student loans. There’s a dynamic where it works really well for you and there are other dynamics where if you don’t know the terms of the deal beforehand, there are different incentives and it doesn’t work out well for you.

Austen Allred [00:19:54] – Right.

Craig Cannon [00:19:55] – Right, so this is kind of, you got burned before with your first company?

Austen Allred [00:19:59] – Yes.

Craig Cannon [00:20:01] – If I remember correctly, you were about to raise an A then it just fell through.

Austen Allred [00:20:07] – We had docs done for an A which is not normally when things fall through. It was very abnormal story. It was December 23rd and we were waiting for a wire any day. Then the gentleman who was leading our Series A, called me and said, “Actually, we’re not.” Well actually he had his secretary call me and say, “Actually, we’re not doing the deal.” And I said, “Can you tell us why?” And like, “Well, we’re just not confident in the company” after a month of due diligence and going through all the paperwork and our law firm was like, we’ve done thousands of these deals, we’ve never seen this happen. We were out of money. I didn’t want people to go through the holidays, planning on coming to work in January and then the company not being there. I called the whole team on December 23rd and said, “I think it’s over guys. I don’t know what other options we have.” We had other people who would’ve funded us, but it’s really difficult to like, “Hey, you know how we went down in a month and a half with that, come with that VC, well, they pulled out so you should fund us.” That’s not, it’s very momentum driven. So that’s why I was very anti VC. Now I think there are companies that should raise VC, there are companies that shouldn’t VC, it’s just dependent on the circumstances and the details and what you’re looking for.

Craig Cannon [00:21:32] – Then you left and you went to go work for another startup right after writing a book.

Austen Allred [00:21:39] – I was broke, where I was working remotely in the middle of nowhere in Utah during this time. We go back home, it’s not like there are tech jobs. The closest tech job was probably two hours away, an hour and a half away. Out of money, I used all my savings to make employees and contractors whole. We were burning it down to the wire with the company funds. And then my daughter who was born, ended up in the hospital for a month. We were just like, actually she was only in there for a week, but it felt like a month. The bills may as well have been for a month. We’re out of money, I’m trying to figure out how to recover and I had written a couple of blog posts about growth as we were working on the company that had taken off and so I said, “Okay, we’re going to turn those blog posts into a book.” I emailed everybody who had subscribed to my blog and said, all right, we’re launching this kickstarter and I’m going to finish up this book. The kickstarter did something like 120000 dollars in sales. That’s what I’m good at, right.

Craig Cannon [00:22:47] – How many people were on the list?

Austen Allred [00:22:48] – A couple thousand.

Craig Cannon [00:22:51] – That’s really good.

Austen Allred [00:22:53] – That, yeah that’s, if there’s one thing I’m good at in life, it’s growing something quickly, building hype for something quickly. That’s kind of my super power to some degree. I used that in its fullest for the first time and then a few companies saw the book and said, “Hey, come work for us.” One of which was LendUp, which is a YC company.

Craig Cannon [00:23:15] – It’s still interesting because it’s fairly risky.

Austen Allred [00:23:22] – Oh yeah.

Craig Cannon [00:23:23] – You lived in the bay before, you lived in your car and then you moved back to Utah. You could conceive of getting a tech job rather than writing your book at that time.

Austen Allred [00:23:34] – For sure.

Craig Cannon [00:23:35] – Why didn’t you do that?

Austen Allred [00:23:40] – I kind of did both. It was whatever I can do to get money right now and a book was honestly faster than a job hunt. Well maybe that’s not true because I could have… It’s actually not true, I don’t know. Okay it’s not a good reason I had offers within days.

Craig Cannon [00:23:59] – I imagine the scenario is like… I’ve been really interested in this idea and question lately of… At what point did you find your voice? People talk about that in the context of art but I think it’s the same for entrepreneurs, in that kind of confidence like, “Oh, I’m just going to do my thing and I could…”

Austen Allred [00:24:22] – Yeah, it’s funny, I had, as everything was winding down, I went to one of our past investors and I was like, “What am I going to do?” I was paying myself, I’d started out paying myself 40000 and I had recently given myself a raise to 60000 and I went to that investor. I was like, “How can I pull this out? I don’t know what I’m going to do next.” He said, “Well, you know, you don’t have any marketable skills, you’re never going to make more than 60000 dollars a year, so maybe you should go find an internship somewhere.” I was like, “Okay.” It’s like that’s where your mindset is right. Maybe I can make 60000 dollars a year. I actually made sure that the first job offer I got was double that just out of spite. But it took me a while to build up the confidence to… People were telling me that, well, when you’re the CEO of a company, you’re only job is to make that company successful. If that company fails by definition, you’re not a good CEO. You could debate whether or not that’s true, but that’s realistically your job is to be successful, right. And so that calls into question, well, what do you actually do? What are you good at? Do you have a skillset? I’d never worked, I’d worked in marketing before, but never in growth. I don’t know if there’s anything I can do.

Austen Allred [00:25:56] – And I didn’t believe that people would trust that I had like a skill set that was valuable because I had investors telling me you don’t. Clearly they know more than I do. It was a psychological, psychologically difficult time and I thought, you know, the only way I can show people what I can do, is by publishing this book.

Craig Cannon [00:26:21] – Right.

Austen Allred [00:26:23] – It’s funny to think about that. I’ve never thought about that, but in retrospect, starting a company is by definition saying I think what I can do is worth more than people will pay me for. Right, I think that I can do something, whether it’s because you can’t measure, I think Paul Graham has an essay about this, something about measurement and leverage and how companies can’t do any of those things very well. If you want to get to a place where you can make a lot of money, get to a place where you can measure it, get to a place where you have leverage that’s why you look at pro athletes, like anybody can measure the number of dollars that Michael Jordan brings in. It makes sense to pay him 100 million dollars a year. Whereas even if you’re the best software engineer in the world at a small company, they’re not going to pay you 100 million dollars a year. Google might, it has in the past spent 100 million dollars to retain Sundar and stuff like that but yeah, it’s mispriced human capital is kind of fundamental to it.

Craig Cannon [00:27:31] – But it’s also a core to your business. But that like, that’s that,–

Austen Allred [00:27:34] – I think that’s the key.

Craig Cannon [00:27:35] – That is the insight.

Austen Allred [00:27:36] – If you look at any asset class in the United States, it’s being traded on Bloomberg terminals instantly all over the world.

Craig Cannon [00:27:44] – Right.

Austen Allred [00:27:45] – You can trade a hundred thousand pounds of pork bellies without anything moving. It’s priced instantly. There are a million people trying to figure out what the accurate prices for those pork bellies. For some reason humans don’t have that, right? It’s you get a job or you don’t. I’m there specifically, there have been women who applied to Lambda School and we get to negotiation stage and they say, “You know, I have always been paid x,” and we have to say, “You’re worth more than two x, so we’re going to pay you what we actually think market is,” and turns out they just always been underpaid and they didn’t know to ask for more. That happens all the time. All the time.

Craig Cannon [00:28:29] – The thing is when you start out, you’re just applying for jobs, right? Take what you can get and then like you mentioned before, many of these, “great jobs” are never advertised, right? You anchor in this weird way that until you know people who make 400 grand a year, you’re just like, well, the job says 75 grand a year.

Austen Allred [00:28:50] – Right.

Craig Cannon [00:28:51] – That’s awesome too. If you’re making 35, 75 is fucking killing it, right? But then you’re just like, “Oh, there’s probably like some ceiling at like a hundred something,” and you’re like, “The answer’s actually no.” Right. It just keeps going. But companies are incentivized to not publish that as well.

Austen Allred [00:29:08] – That’s the funny thing when you get to Silicon Valley and you see there is a desperation around top talent and there are people commanding million dollar salaries and the companies will happily pay it. Obviously that’s not an entry level salary, and there are people in the Midwest who are capable of being paid six figures in the Midwest that are making 20 in the Midwest and that’s crazy to me. That’s the problem we’re really out to solve is why is there no mechanism to help people realize their full potential, both in economic standpoint and a non economic standpoint. And why is there like there’s nowhere near an efficient marketplace for capital. Nowhere close and that’s crazy because if you look at GDP and spending and like where we can make the most improvement as far as efficiency goes, human capital is the most unoptimized asset class we’ve ever had.

Austen Allred [00:30:13] – And that’s insane.

Craig Cannon [00:30:14] – Well, especially in the states too where people aren’t working in factories for the most part.

Austen Allred [00:30:18] – Right.

Craig Cannon [00:30:19] – Aside from the tech jobs that you offer right now, what other opportunities do you see as like really great for arbitrage in that way?

Austen Allred [00:30:29] – Right now we’re hiring an economist to help us make sense of all that. It’s something that you don’t have to be an economist to see. You can download data from the US Department of Labor and see like how many people are applying for the jobs and how many jobs are out there. There are some that have drastic, like three to one shortages that pay really well and there just aren’t enough people. Nursing is one of them. Every nursing school is at capacity because they have some. There’s something called the 90-10 rule, which says that only 90% of your money as a school can come from federal funds. The other 10% has to come through people paying tuition or private sources somehow. That is actually the thing that drives the education system in the United States. That 10%, finding that 10%. Every school is just desperately trying to find that 10% and however big that 10% is, they can blow the rest out with federal funds all day. Nursing schools, for example, there’s a huge shortage of nurses. Boomers are getting old, millennials aren’t going into medicine, but nursing schools are capped because they can’t find that 10%. That drives massive labor shortages across the entire economy and that’s crazy. That’s just a market inefficiency. We’re not working on it right right now, but don’t be surprised if you hear by the end of this year an announcement that Lambda Nursing is going live and we’re launching in nursing school or you know, so we look at it backwards

Austen Allred [00:32:16] – from a labor market perspective and we’re starting out with where the shortages in labor markets and let’s fill those gaps and turns out some of those gaps are five million people at a time. That’s a huge inefficiency and if we can be a toll road that takes a very small toll to get you there, it’s an enormous company.

Craig Cannon [00:32:36] – I think so too. I’m curious about then how you end up modeling that risk if the program, I imagine the nursing program might be a little bit longer than nine months. You’re going to have to give people, I imagine like you’re talking about testing out these living stipends, stuff like that. How do you start modeling that risk?

Austen Allred [00:33:00] – In the beginning you make good guesses and you put it on a spreadsheet and you give yourself a margin of error. It looks like any other fintech industry where you start out saying, “All right, we’re going to make some loans. We don’t really know how it’s going to work out. We’re going to learn from that and over time we’ll have a really predictive model.” We’re getting better on the just income share agreement for education side for software engineering. We’re not all the way there, but we to some degree, know what, can predict what the returns will be given the macro economic environment staying the same. You kind of just have to like create terms that make sense and figure it out as you go. When we talk about a moat, over time, you can better predict what repayment will be. You can better predict how much money you’re going to make. You can write a tighter model. If you can write a tighter model, you can bring down the cost of capital, you can bring down your cost or you can increase the services that students get. It is just a fundamental advantage to having done that before

Austen Allred [00:34:04] – and now we’ve got a couple of years head start on everybody else and we can pretty act, not with 100% accuracy, but we have a pretty good sense of how successful a student will be by the time they’re starting or a couple months on.

Craig Cannon [00:34:17] – Wow.

Austen Allred [00:34:18] – We’re hiring teams that are going to focus on just that. How can we shorten that feedback loop so that we can determine very quickly how successful people will be.

Craig Cannon [00:34:30] – Can you quantify someone’s work ethic or grit?

Austen Allred [00:34:35] – No, but they all along the way we’ll spit off data that you can use to try to approach that. The numbers are an output, not input. Similarly to a startup, right. You can’t quantify the grit of founders, but you can look at what they’re doing and the results of that you could retroactively say, “Yeah, that looks gritty.” If you get enough data points, machine learning is actually… That’s one of the places where AI makes the most sense is saying, “Let’s take all of this data and let’s figure out a model that maybe humans couldn’t put together, but computers can figure it out.” Pretty soon you can underwrite the likelihood that a person will be a successful software engineer before they’ve paid a dollar. And then that changes everything.

Craig Cannon [00:35:29] – Someone asked this question so I want to credit them somewhere. I know who it was, yeah, David Kafeidwind basically asked, “Imagine a market dynamic where it’s mostly white dudes getting hired for software engineering jobs. Is it possible that your model could learn in the wrong way and say, ‘Only select white dudes for Lambda School?'”

Austen Allred [00:35:51] – It could, if that’s what we’re actually happening. That’s not what’s actually happening. The reality is if we can make someone a strong software engineer, they’re getting hired anywhere but there are ethical concerns that we have to be aware of right. I used to work in lending. You have to be very cognizant of that in lending because for example, if you use zip code and you say, look, the people from this zip code tend not to repay, zip code is very directly correlated with demographics and their ethical concerns that you have to be wise about. What we’re seeing right now is actually the reverse to some degree. Obviously not, it’s not causation by any stretch of the imagination, but we can tap into markets of completely untapped talent because we’re not charging anything. Folks who are brilliant just to haven’t had opportunity before and so we actually overindex to those types of communities. As a school we’re less than half white, which is not the traditional makeup of a code school by any stretch of the imagination. But that’s not because we’re intentionally skewing admissions that way, it’s just because that’s where a lot of untapped talent is. We try to find and train that untapped talent and companies are happy to have it. In fact, it’s a huge need for companies. We don’t sell ourselves as a diversity solution, but it just so happens that our students who are great engineers are generally speaking more diverse than Stanford would be.

Craig Cannon [00:37:39] – Yeah, that makes sense.

Austen Allred [00:37:42] – But great question.

Craig Cannon [00:37:43] – Yeah, it’s really interesting. You guys now are remote, right? All the coursework happens live stream. Do you see a version where you become centralized? My buddy just, he worked at the college in Vermont that just went under. I imagine you can buy colleges at this point. Do you see a world where maybe, probably with the nursing program you’ll have to, you would have to buy.

Austen Allred [00:38:12] – It changes the cost structure. The less we can be in person, the better. We have some experiments that we’re running where we were actually housing people in group housing together. They live and work in the same environment. They’re still taught online because we don’t have to have the density of instructors in place at one time, we can just get students together. I’ve looked at, “Hey, can we just buy a summer camp and run Camp Lambda and have lunch?” Like that’d be awesome. We just have to decentralize housing is much cheaper than centralized. Students would have to be willing to build that into the cost. So what I like to say is, you look at Amazon, Amazon in the beginning was saying, no, we’re entirely online and we’re going to be focused there. But once that’s nailed, then you start to play with other stuff, we’ll probably be the same way. But the core of Lambda School, for the foreseeable future is online whether students are in the same place or not. We actually have a monthly meetup in every major city in the US.

Austen Allred [00:39:24] – Where students all get together, in person and that’s working pretty well. There may be more of that, especially as we have more density. but the educational experience will probably always be online.

Craig Cannon [00:39:37] – This is what I’ve heard from the guys at Zapier who were just saying, you basically have to be intentional and like build that remote culture.

Austen Allred [00:39:45] – Very intentional.

Craig Cannon [00:39:46] – Because if you don’t, you could easily see a version where, “Oh, you weren’t in class today,” then you don’t hear the thing, you didn’t hear the tip right.

Austen Allred [00:39:54] – And I think you have to do that as a company, as a school. It’s not something that happens on accident. In the beginning, we started out by figuring out, okay, here are all these people who could be watching Netflix right now. It actually goes back to our instructional design where, I mean, if I think back to my college classes, maybe 50% of people were paying attention. If I’m being optimistic, at least… This was, call it 2010, at least 50% of the class was on Facebook.

Craig Cannon [00:40:23] – Have you seen those photos from back in the day when Facebook just started?. I don’t think it’s Photoshopped.

Austen Allred [00:40:30] – But I imagine there’s a real one out there somewhere where it’s like from the back of the class and the professor’s talking and like it’s 70% of the screens are Facebook. Oh yeah. For better or worse, I actually think it’s a secret advantage is that we can’t rely on the crutch of just like looking at somebody and seeing what they’re doing. We have to use data and software and instructional design to make sure everybody’s working on the right thing at the right time. It’s a different world that you have to build for and you have to build it for it from first principles. You can’t just tack that on at the end, “Oh, by the way, we’re online.” That does not work well.

Craig Cannon [00:41:09] – Well, most MOOCs are yeah disasters.

Austen Allred [00:41:11] – Correct.

Craig Cannon [00:41:12] – 10% or something.

Austen Allred [00:41:13] – Of the good ones.

Craig Cannon [00:41:15] – Yeah, it’s crazy. Dave Dawson asked several good questions. He says, “you appear to be on the successful path now,” we’ll just take that for granted.

Austen Allred [00:41:26] – Sure.

Craig Cannon [00:41:27] – “Was there a point early on in Lambda School when you want it to stop?”

Austen Allred [00:41:34] – There was never a point I wanted to stop. If you paid me minimum wage to do this forever, I would do this forever for minimum wage. Maybe I’m just a sadist or masochist or something .

Craig Cannon [00:41:51] – You might have to move back to Utah.

Austen Allred [00:41:56] – But there are definitely times when it was just like, this is not going to work and we’re screwed and in the beginning it was like all those other schools were right. Like when we started, everybody said, A, you can’t teach people effectively online. B, you can’t teach people effectively without skin in the game, they have to put something up. People aren’t going to get jobs if they don’t have a network. You can’t accept the categories of people that you’re accepting and have them be successful software engineers. Which looking back like a ridiculous thing to say but everybody believes that. Like the entire code bootcamp industry believes that online was not very effective. We have an in person class and like, “Yeah, let’s throw it online,” and “Oh, it didn’t work.” Well clearly that’s fundamentally broken. Everybody who had ever tried a free upfront thing didn’t use any sort of filtering other than a paper resume and people would show up on day one and they wouldn’t show up on day two and they’d say, “Oh, that’s clearly isn’t working.” There are a whole lot of times when we said this, “Maybe they were right.” That was the real reaction. Maybe we were wrong this whole time and we’re out here saying, “Hey, you guys are all old school. You’re all doing it the wrong way. We’re going to do it over here.”

Austen Allred [00:43:23] – Well, maybe there’s a reason that hundreds of years of education have worked this way. But really I think what happens is we don’t have another option. We have to figure it out or we die. That’s a very powerful motivator.

Craig Cannon [00:43:40] – Not getting a real job. Have you ever completed a online course before?

Austen Allred [00:43:47] – I have not. And in fact, we were talking about the book. The book was available as an online download and one of the things that made me want to build in this model it and I sold the book for 100 bucks. It was not cheap. 50% of the people that bought the book never downloaded it even. Between looking at that and looking at MOOCs, it was like, has anybody ever figured out how to make education actually successful? Or are we all just opening the top of the funnel so wide that some people are going to fall out the bottom and that’s become my thesis, is that there is some percentage of society who are autodidacts and they just have a ridiculous advantage over everybody else. Part of Lambda School’s how can we help the other 98% learn how to be autodidactic? How to see a problem, approach it, figuring out a solution? We approached instructional design with no assumptions and I think every school ever has looked at everything and said, “Okay, well here are the successful students so our job is just to weed everybody else out.”

Austen Allred [00:45:05] – When really like why are those, why are the other students unsuccessful? Is that because they suck or is that because you suck? And we’ve found that it’s really just because you suck most of the time. And it’s really easy for schools to just look the other way.

Craig Cannon [00:45:24] – I have been wondering, I’m doing a Duolingo right now and I’ve been curious about, I mean, it’s cool, but it’s very open. You can just choose where you want to go, they’re like motivate you a little bit by keeping a streak up, but I’m actually a little surprised at people, if people finish it because I’m doing it every day.

Austen Allred [00:45:43] – I think if you guess what their completion numbers were, you’d be right. I think about, you know, when I went to college and I had a class and I showed up and it was a Shakespeare class, I think about the work that I put in in that class and then the work that I put in outside of that class, there’s no way in hell I put in as much work studying Shakespeare on my own as I would in the four walls of a university. But I don’t think that’s because there was a physical campus with a professor wearing a tweed jacket necessarily, that’s just because there, there was social pressure, there is instructional design, there were external mechanisms that force you to put in the work. We had to like, that’s what we built into Lambda School.

Craig Cannon [00:46:27] – But you also have to make it fun.

Austen Allred [00:46:29] – Totally.

Craig Cannon [00:46:30] – You have to motivate in that way you know like it’s why I learned to program, it’s why I learned to video edit, it’s why I learned everything on the side because you had a clear goal like, “Hey, I want to build this thing and it’s enjoyable to build the thing you want to build.”

Austen Allred [00:46:43] – You look at as an example, computer science programs, right. They have ridiculous dropout. And they’re proud of that but in my opinion, it’s because they start like writing Java or they start writing C and you have to go semesters before you get to a point where you can write any code that you can see or touch or actually play with. Yes, you’re getting the fundamentals but Lambda School works the other way. You have a site that you can play with the first day and it’s not… You don’t understand the fundamentals of programming. It’s crappy html and crappy CSS, but you can see it, you can touch it, you can play with it. And I think there are a lot of people, if not the majority of people that learn that way. And then the people who aren’t going to become researchers and that’s okay and just because you start at that point doesn’t mean you can’t drill down until you’re doing the fundamentals. You do need to know how to write C to be a good engineer. You do need to know how memory management works, but maybe let’s not start with memory management as like an entry level point to software engineering.

Craig Cannon [00:47:48] – I did it a year of CS and even high school and it was just like, what are we doing this for?

Austen Allred [00:47:54] – Right.

Craig Cannon [00:47:55] – What’s the is the point of this?

Austen Allred [00:47:56] – And there are a few people that are interested by that enough that they can carry on, but that’s not enough people to build the world of software that we need.

Craig Cannon [00:48:08] – Dave asked another question that I want to cover and that is, “What keeps you up at night now?”

Austen Allred [00:48:16] – We agreed as a team to no longer share the number of students that we have enrolled. But it is not a small number and it is growing at an insane pace. We’ll soon be measuring Lambda School’s scale by percentage of the overall number of students learning to program every year to give you some context and being able to support that kind of scale, effectively is what keeps me up at night. Mostly specifically hiring, mostly hiring executives right now, who can build out 100, 150 person teams beneath them. Those people are difficult to find and it’s a very important decision and we need like five of them yesterday. On the individual contributor side, we’re probably, we’d need to hire like a hundred people very quickly and that is hard.

Craig Cannon [00:49:28] – But that would be remote, right, those jobs, Yeah, so if you’re looking for a job.

Austen Allred [00:49:34] – Yeah, oh please. I think we’re hiring for like 50 positions right now. lambdaschool.com/careers. Please, please, please.

Craig Cannon [00:49:42] – All right, so I’m going to talk about like two other topics. One, actually really like this one. Dayo Koleowo asks, “I’ve made remarks I do not agree with.” That’s a quote from your Twitter bio. Yeah, you’re a prolific on Twitter if people don’t follow you. “What is that one remark you wish you didn’t have to disagree with?”

Austen Allred [00:50:04] – One remark that I–

Craig Cannon [00:50:05] – What he’s trying to get you to say is what is something that you may have had to like, say, or go back on a perhaps like politically correct or like environmentally correct.

Austen Allred [00:50:21] – A lot of layers to that question. That’s a tricky one.

Craig Cannon [00:50:24] – Or maybe even a core belief where you’re just like, “Oh, maybe I’m just wrong here.”

Austen Allred [00:50:29] – The one that I go back and forth on the most, is as a company being remote versus in person and the trade offs of that. We bring people together a lot and I see the advantages of that and a lot of our executive team will probably end up being in person. And so it’s, I would say that I wish that remote work, were at this incredible place where there’s no trade off. I think that’s something that everybody wants to believe and some people do believe. But until you’re kind of in the CEO or executive role trying to bring a company together, you don’t see some of the trade offs. I would love to be able to move anywhere and not pay what I pay for rent and by extension not pay everybody else, but I have to pay them so that they can pay rent.

Austen Allred [00:51:37] – But I still think and a lot of our students come in wanting to be remote. A, I don’t think remote work is built into enough companies yet where I would be comfortable telling somebody that I love that they should start their career as a remote junior software engineer. B, I think we’re still figuring out what the right dynamic is of in person versus remote, like us as a tech industry, what the trade offs are and what the ideal way to do that is. My first company was entirely remote and that’s really hard when you’re trying to come together and create something, but nobody’s in the same spot. They’re just trade offs. That said, all of our instructors are remote. All of our career coaches are remote and all of our students are remote and because we’ve built around that, or we’ve built into that it doesn’t matter. The two things that are most like my most controversial beliefs maybe or most like seemingly like don’t seem to jive with each other, is that I think the best high growth companies are still going to be built in Silicon Valley for right now or at least they’ll have an executive team in Silicon Valley, but also I believe the best schools are remote. I don’t think that’s a contradiction though it certainly feels like one.

Craig Cannon [00:53:20] – That’s a good way of thinking about it. Yeah, I was listening to Michael Seibel talk yesterday and he said he’s had his mind changed about remote companies given the success of Gitlab, Zapier type companies, right. But the way he puts it now is that it has to be one of your like core things, one of your core innovations and if you’re thinking like maybe we have the energy or the insight to innovate on a couple, maybe three categories, are you willing for remote work to be one of them?

Austen Allred [00:53:52] – I think that’s exactly right. You can only solve so many problems as a company and I mean, I’ve talked to the CEOs of both Zapier and Gitlab in the past couple of weeks and they think about that all the time. And the nice thing is the talent advantages that you have when you do that are enormous. When Zapier’s like, “Yeah, we’ve got like this guy in Ohio and this guy in Nebraska.” It’s like, man, you know, I’m hiring but at the same time that the trade offs are that if you have an office in the financial district and you’re hiring five executives, there are a hundred companies that have been high growth that, you know, the funny thing is a lot of the people who they join a high growth company, they know how to grow it, they figured it out now they’re successful and they have less incentive to move than other people do.

Austen Allred [00:54:47] – The pool of high growth executive talent with 15 years of experience growing high growth companies is still largely concentrated in San Francisco. If you need or call it Silicon Valley.

Craig Cannon [00:55:00] – Whatever.

Austen Allred [00:55:01] – If you need 50 people to do that, it’s really difficult to find 50 people in Ohio to do that. It’s still difficult to find 50 people anywhere, but if you’ve built that in, like Gitlab and Zapier have and ther companies like InVision or whatever else, they’ve chosen that as their path and they have to spend a whole lot of time thinking about and focusing on and optimizing for that. Maybe that’s the trade off where they have to spend less time hiring or less. There’s a less competitive hiring market. The instructors that we have now, there’s no way on earth we could have had a similar quality of instructor base in San Francisco. They’re just, there’s fundamentally is not. It’s a trade off. It’s difficult.

Craig Cannon [00:55:54] – That’s a tricky one. Okay, so, the last thing I want to cover is I’ve heard you mentioned a couple of interviews that the market is not very good at measuring product quality.

Austen Allred [00:56:04] – Correct.

Craig Cannon [00:56:05] – Or something to that effect.

Austen Allred [00:56:06] – What I usually say is specifically analysts. If you look at public companies or VCs or whoever else and another way to say this, I think PG has said that revenue is a very, very trailing indicator of product.

Craig Cannon [00:56:28] – Right.

Austen Allred [00:56:29] – Back when I had spare time, I would play the stock market a lot and people just don’t have a very good, there’s no way you can quantify product quality very well. People who are looking at things from a quantitative standpoint, underestimated every single time. I look at some of the classic companies like look at the YC companies, the Airbnbs, the Stripes, they all had competitors. Some of their competitors had a headstart, some of their competitors had more money and more market share at different times, but there’s a reason those companies win and become durable companies. Another, Dropbox is an example.

Craig Cannon [00:57:17] – Well it’s one of my favorite Steve jobs videos. I don’t know if you’ve seen this one before, but it’s like a very hot burn on Microsoft and where he’s just like, “Listen, they’re good people. The problem is they just have no taste.” What I was curious about with you is, do you think you can teach taste in a school? Because it’s very much, it’s related to product, it’s related to listening to users, it’s all of that. But there’s something more ethereal about it.

Austen Allred [00:57:47] – The main thing is having thought about it a lot and then having seen it and having noticed it when you see it. For example, we use Slack as a school. We’ve got, I could tell you how much we spend that would tell you how many students we have. It’s a nontrivial amount of money for Slack. And people say, “Why don’t you use, you know, competitors or you know, just use IRC,” but it’s very, very subtle. The work that has been put into Slack and Slack it’s not perfect. But as a product you can just tell. It’s a whole lot of the way I like to think about product is a whole lot of subtlety built up over time to the point where it becomes an advantage. If you build up a lot of those advantages over time, it compounds to some degree because it’s difficult to measure and it’s so subtle. For example, if you drive a Model S and then you go drive a similar cost Jaguar for example, the Model S like the materials are worse, the build is probably not as good. They’re not as good at assembling the pieces, but just the experience of that car versus the other one it’s not even close.

Austen Allred [00:59:18] – Tesla is probably the canonical example of product matters way more than analysts can measure.

Craig Cannon [00:59:26] – Cool, so you just think expose students to this over a long enough timeframe.

Austen Allred [00:59:31] – When I started to realize it is when I had a few product minded people point out some of the subtle details and be like, you read this, like you find yourself looking for one app or one piece of software or using something more often. And it takes somebody to sit down and point out, “You know why you’re doing that, right? It’s because this is happening at this specific time in this specific way.” And you’re like, “Oh, that was very pleasant actually in retrospect.” Pointing that out and then beginning to look for it is when I think you start to notice the key differences.

Craig Cannon [01:00:07] – Fair enough. I like being critical sometimes. Especially of products. All right man, so if people want to sign up or learn more about Lambda School, where should they go?

Austen Allred [01:00:16] – Yeah, at lambdaschool.com.

Craig Cannon [01:00:18] – Sweet and you’re on Twitter at?

Austen Allred [01:00:19] – @Austen with an “e.”

Craig Cannon [01:00:20] – Cool.

Austen Allred [01:00:21] – Like the feminist author, not like the city.

Craig Cannon [01:00:25] – All right, thanks for coming in.

Austen Allred [01:00:25] – Perfect, thank you.

Craig Cannon [01:00:27] – All right, thanks for listening. As always, you can find the transcript and video at blog.ycombinator.com. If you have a second, it would be awesome to give us a rating and review wherever you find your podcasts. See you next time.


  • Y Combinator

    Y Combinator created a new model for funding early stage startups. Twice a year we invest a small amount of money ($150k) in a large number of startups (recently 200). The startups move to Silicon