SDS 535: How to Found, Grow, and Sell a Data Science Start-up

Podcast Guest: Austin Ogilvie

December 28, 2021

In this episode, Austin describes his journey from the arts to programming, bootstrapping a startup, his take on what makes a great data scientist, his everyday tools, his practical tools for aspiring tech founders, and more!

About Austin Ogilvie
Austin Ogilvie is the CEO of Laika, a compliance as-a-service platform. He is the Former CEO of Yhat, a data science platform acquired by Alteryx.
Overview
Yhat, Austin’s company, is a huge content creator, Jon shares a time when he would put in data-related queries that would often return a Yhat blog post. Austin and his co-founder started as product managers at a capital company during an interesting time for the open-source community as many languages were leaving paid platforms. Austin and Greg became interested in sharing real-world applications of these tools in a world of lacking demos. So, they started as personal bloggers until founding the company and migrating content there which attracted the attention of data scientists as well as users and investors.
In 2013 the company was founded and they found a way to cram seven people into a single apartment. Austin and Greg found in their work as product managers needed intense and expensive predictive models. So they wanted to transform this work from something academic and into a user experience on mobile. Yhat provides an end-to-end platform for developing, deploying, and then managing APIs, removing the usual blocks and pain points. It wasn’t flawless though, since Austin and Greg are both product people. They found the production roadblocks frustrating, preventing them from getting their product to consumers. At the time, a data science team was a rare thing, as well. But Yhat was in the Y Combinator program which provided funding, mentorship, and networking. Ultimately, Yhat was acquired in 2018. Austin had both an exciting and challenging journey to go from their 17-person startup to what exists today.
What’s interesting about Austin is he started out studying Arabic and Spanish. In undergraduate, he was interested in travel and politics (something he got to explore after the acquisition of his company) but got into technology during an internship at a startup in D.C where he “caught the entrepreneurial bug.” He sought out more experiences in young tech companies and found himself interested but frustrated that he didn’t understand the technical aspects. So, he began studying product design and building in data science. He attributes this ability to shift careers to the privileged time we live in where access to information and learning is vast.
Following the acquisition by Alteryx, and some well-deserved time off, Austin founded Laika, a tech start-up that solves the issue of compliance– a problem that came about during the merger and acquisition process with Yhat. There is a significant growth lever in expediting the compliance process, and Laika does just that. It helps expedite deals, trust with partners and customers, and may help companies enter new markets faster. Essentially, it is a growth machine that provides the peace of mind Austin would have wanted at Yhat.
Beyond his work at Laika and Yhat, Austin is also interested in gaming environments, specifically Unity. As Jon points out, Unity is a huge tool for deep reinforcement learning. Jon suggests deploying deep reinforcement learning algorithms into a game engine, specifically Unity for 3D environments.
We concluded the lively hour with Austin sharing his top tips on becoming a successful technical founder. As a former liberal arts graduate who managed to accomplish what many have tried, but only few have achieved, Austin shed light on some universal advice that we’re sure you will be taking note of. He stresses to be generous with your time; be honest about needing help and asking for it; have faith in yourself and know that you can achieve your dreams; and finally, if you’re dreaming of starting your own venture, Austin recommends working at an early-stage tech start-up to gain exposure to processes that will help you grow quickly.
Whether you’re a data scientist who plans on building a new venture or a tech enthusiast just getting into the game, we hope this conversation inspired your inner entrepreneur to seize the day and get started on your dreams this year.

In this episode you will learn:
  • The story behind the naming of Yhat and its early beginnings [5:10]
  • Austin and Yhat’s experience at Y Combinator [19:00]
  • The benefits of being a technical founder [25:00]
  • From arts degree graduate to successful tech entrepreneur [27:00]
  • Austin’s latest venture, Laika [39:30]
  • The tools that Austin uses day-to-day [47:30]
  • Unity gaming environment [49:58]
  • What makes a great data scientist [56:23] 
     
Items mentioned in this podcast:
Follow Austin:
Follow Jon:
Episode Transcript

Podcast Transcript

Jon Krohn: 00:00:00

This is episode number 535 with Austin Ogilvie, founder and CEO of Yhat and founder and CEO of Laika. 
Jon Krohn: 00:00:13
Welcome to the SuperDataScience podcast. My name is Jon Krohn, a chief data scientist and bestselling author on Deep Learning. Each week, we bring you inspiring people and ideas to help you build a successful career in data science. Thanks for being here today and now let’s make the complex simple. 
Jon Krohn: 00:00:42
Welcome back to the SuperDataScience podcast. Today’s guest is the rapier-witted Austin Ogilvie, a prodigiously successful data science entrepreneur. Austin was the founder and CEO of Yhat, a Y Combinator backed startup, that built tools for data scientists and became a brand with a cult following in the data science community. In 2018, Yhat was acquired by Alteryx, an analytics automation company that’s publicly listed on the New York Stock Exchange. After Yhat, Austin became a serial entrepreneur when he founded Laika, a compliance as a service company, that dramatically improves your capacity to sell your product. Laika last month closed a $35 million Series B funding round. Only two years ago, that brings the total funding raised by the firm to a staggering $48 million. In this episode, Austin describes his journey from an arts degree, studying foreign languages, to teaching himself programming and machine learning, and then bootstrapping a data science startup into a respected brand and acquisition target. 
Jon Krohn: 00:01:46
He talks about his unique take on what makes a great data scientist. He talks about the hands-on data science tools he finds great value in coding with day to day, even as the founder and CEO of fast-rowing tech startups. And Austin covers his straightforward and practical tips for growing into a successful technical founder yourself, regardless of whether you yourself have a technical background today or not. Today’s episode will be of great interest to anyone who’s interested in founding, growing and/or successfully exiting a tech startup, particularly if you’re thinking of incorporating data or artificial elements into your startup. We do also discuss technical data science tools here and there, but largely only at a high level. All right. You ready for this inspiring episode? Let’s do it. 
Jon Krohn: 00:02:42
Austin. 
Austin Ogilvie: 00:02:43
Yes. 
Jon Krohn: 00:02:44
Welcome to the SuperDataScience podcast. 
Austin Ogilvie: 00:02:46
Thank you having me. I’m excited. 
Jon Krohn: 00:02:48
Yeah. I’m thrilled to be able to reconnect with you here in front of this audience. I’ve known you for a really long time. So we were actually introduced, it took a while to figure this out, but it seems like 2013, 2014, Sam Hett, who worked at meta markets at the time, which was an ad tech company, and I was in the ad tech space, we met and I think he literally said something to me like, “I’ve got a friend Austin who’s really interesting that you should meet.” 
Austin Ogilvie: 00:03:20
It sounds about right. It definitely was that far back for sure. And I’m very glad that you and Sam are both out of the ad tech game. 
Jon Krohn: 00:03:30
Well, yeah, I mean, I definitely, I like where I am now with what I’m doing. I’m not going to just bad mouth the ad tech space on air unilaterally. But yeah, I think that there are interesting problems that can be solved in other spaces too.
Austin Ogilvie: 00:03:43
Totally. 
Jon Krohn: 00:03:45
And then more recently we were reintroduced by Drew Conway who was on episode number 511. And that was the first episode that I ever did live in front of an audience, which was really exciting. 
Austin Ogilvie: 00:04:00
Yeah. That was a good one. Yeah. And Drew’s been a mutual friend of both of ours for probably just as long back to 2013, at least. Right? 
Jon Krohn: 00:04:11
I’ve never met Drew in person. 
Austin Ogilvie: 00:04:12
No way. 
Jon Krohn: 00:04:12
No, I was introduced to him with a purpose of having him on the podcast. 
Austin Ogilvie: 00:04:17
Amazing. That’s such a cool connection. 
Jon Krohn: 00:04:21
It really is because he’s somebody, his book, Machine Learning For Hackers, that was a key book for me, transitioning from academia into professional data science. 
Austin Ogilvie: 00:04:32
Yeah, me too. It was that one and Matthew Russell’s, Mining the Social Web, which was a Python book, very similarly spirited to Drew’s and John Miles White, right? 
Jon Krohn: 00:04:44
Yeah. John Miles White was the other author on there. 
Austin Ogilvie: 00:04:47
Yeah. So when we met, I was running my prior startup called Yhat and it was a data science platform. I’m sure that Sam put us in touch because we were both nerding out about the same stuff at the same time. 
Jon Krohn: 00:05:01
I was also potentially a client. 
Austin Ogilvie: 00:05:03
Probably that was an angle. Yeah. 
Jon Krohn: 00:05:07
So yeah. So the company name is something that I’d like to talk about first. So Yhat, brilliant name. So many listeners, people who are familiar with machine learning statistics, will already be aware of this, but we typically use the variable Y to represent the outcome that we’re trying to predict with a model. And when it has the little hat on top, circumflex, it’s this two-sided triangle that is on the top of a Y. And it’s the actual prediction that a machine learning model makes in a given circumstance. And so absolutely brilliant name for a data science company. I wish I thought of it first. 
Austin Ogilvie: 00:05:52
It was a really fun name, but the list of pronunciations that we would get from outsiders, non-data nerds was hilarious. A lot of like, “Is it pronounced yacht or what is this?” I always assured people that our customers understood. 
Jon Krohn: 00:06:09
I mean, that’s the key piece of it is that it’s like a secret handshake. It’s like you know right off the bat, if somebody can pronounce your company name right, then they’re in your target market. 
Austin Ogilvie: 00:06:19
That’s right. 
Jon Krohn: 00:06:20
If they can’t, then it’s probably a waste of your time and you can end the call. 
Austin Ogilvie: 00:06:22
Yep. In fact, we trained all of our salespeople. That was the qualification standard. 
Jon Krohn: 00:06:28
Actually? 
Austin Ogilvie: 00:06:29
No. 
Jon Krohn: 00:06:29
No. You have this big juicy contract but you couldn’t pronounce our name so we’re cutting you. So yeah. So amazing company name, but that’s not all. Also, an amazing company. So there were, at that time, almost, not almost any query, but I could write a lot of questions into my Google search related to data science. And very frequently the answer that come up was a Yhat blog post. So you guys did a ton of sharing with the community of Python tricks, R tricks, stats tricks, machine learning tricks. You’d publish them and write them in a really compelling way, You must have had a lot of your blog posts that must have been high up on hacker news I’m sure. 
Austin Ogilvie: 00:07:19
Yeah, totally. Greg, my co-founder and I, he and I were both product managers before we founded Yhat at a company called On Deck Capital. And we started there. I started in 2010. Greg joined about a year later. And it was a very interesting time for the Open Source statistical programming community. There was a big tailwind behind R and Python. Lots of organizations had been recently transitioning off of paid languages like MATLAB or SaaS. And the Open Source tools that were available were really coming into an entirely new level like Pandas was probably first released in earlier 2007, 2008, I would imagine. But by 2010 it was really quite a robust platform. In any case, long story short, Greg and I were big users of these tools, scikit-learn Greg had been using very early in that project’s life. Pandas, the various R packages, the Tidyverse, et cetera. And we were just really interested to share real world examples in a sea of very toyed demos that we thought were out there. A lot of like Iris data sets and so forth. And we were really interested to share what we were doing first at work at On Deck before we started Yhat. We got into this and blogging on our own personal blogs first. And then when we decided to start the company, it was just very natural for us to migrate all the content there and to continue that tradition. 
Jon Krohn: 00:09:05
Yeah. And it definitely worked. It paid off. I’m sure you were just doing it out of interest, like to engage people with this content, but in terms of a brand strategy, it just coincidentally worked out incredibly because you developed this cache as leaders to look to for any of these kinds of questions. I knew that if one of the top results for a data science question that I had was a Yhat article, it was going to be well written. It was going to be clear. And so, yeah, so I’m sure that that was great for hiring and for getting business and anything like that. 
Austin Ogilvie: 00:09:47
It was really fun to build and definitely very influential. At peak, I think we had like 100,000 or so unique visitors a month, which nobody sees that, but the data science community is growing, fast growing, but it isn’t that big. 
Jon Krohn: 00:10:04
Yeah, exactly. If you have a general website for anybody that has a million visitors a month, that’s huge. And so to have 100,000 when only 1% or fewer … it’s got to be way less than 1% of people are data scientists or people who are looking at coding problems online. So that’s like the equivalent of having a 10 million view website when you adjust for the market size. So super cool. The products that you were building were really practical as well for data scientists. So I remember, I think maybe the first call that we ever had back in 2013 or 2014, was you giving me a demo of the rodeo Python IDE, the integrated development environment. So the idea of an IDE, like RStudio people might be familiar with, is to bring lots of different kinds of functionality that could be in different panels altogether into one view. So you can see your variables, you can see the code you’re executing, you can see your outputs, you can see all your file names. And so I guess actually, maybe that’s a little bit of an analogy. It’s like an RStudio for Python. 
Austin Ogilvie: 00:11:15
That was our whole objective. We were obsessed with RStudio, huge RStudio fans. And at the time there was no support for Python that was remotely close to that experience. So we built rodeos like a pixel for pixel sort of clone of RStudio. We obviously didn’t get nearly as far into that project as RStudio has spent many decades at this point working on. But yeah, that was the goal. There are other solutions that have come about that are way better now. VS code has a lot of support for Python and data science specifically. But yeah, at the time that was a really, really fun project that had a lot of early legs. 
Jon Krohn: 00:12:04
Yeah. And so it attracted the attention of not only data scientists as users, but these cool tools that you were building attracted amazing investment. So you guys were in the Y Combinator program. 
Austin Ogilvie: 00:12:20
Yeah. So we founded the company in 2013 in New York and we did $1 million seed round. We were roommates working out of our apartment for the first year or so until we got to be a team of seven. And then we finally got a basic office. 
Jon Krohn: 00:12:42
You were in the apartment up to a team of seven? 
Austin Ogilvie: 00:12:44
Yeah. It was really inappropriate. Yeah. Yeah. Just in Carol Gardens on Court Street. Everyone would show up in the morning and there we were. 
Jon Krohn: 00:12:56
Making a cup of coffee. 
Austin Ogilvie: 00:12:57
Yep. We grew out of that eventually, but yeah. So Greg and I have been working at On Deck, which is an alternative lending company, sort of like Lending club, but for small businesses. And we were both product managers. So our charge was to build stuff that customers would use, but a lot of the underlying functionality that the web apps we were building needed was advanced risk models and dynamic prices, risk-based pricing, credit scores, this type of stuff. It was a very, very painful process by which a data science team would express whatever model or imputation strategy they wanted to deploy into production. And that was the impetus that precipitated our leaving to go start Yhat was like how do we take this very expensive and complex and valuable data science work and transform it from something simply academic that is on a laptop into an iPhone experience or a web app experience. And now there are some wonderful tools for this, but at the time, our strategy was very bad. We would just rewrite the code into, we were using a lot of Java, a lot of Ruby and little details like R has a different floating point decimal precision than does Java, these little details are very consequential. 
Jon Krohn: 00:14:31
Zero indexing versus one indexing. 
Austin Ogilvie: 00:14:33
Exactly. 
Jon Krohn: 00:14:34
[crosstalk 00:14:34] confusion. 
Austin Ogilvie: 00:14:35
What kind of a sick person indexes at one? My gosh. 
Jon Krohn: 00:14:39
Yeah. I know. But somehow the entire R language- 
Austin Ogilvie: 00:14:42
The whole thing. 
Jon Krohn: 00:14:43
I know. Statisticians, I guess. 
Austin Ogilvie: 00:14:46
Yeah. 
Jon Krohn: 00:14:54
Eliminating unnecessary distractions is one of the central principles of my lifestyle. As such, I only subscribe to a handful of email newsletters, those that provide a massive signal to noise ratio. One of the very few that meet my strict criterion is the data science insider. If you weren’t aware of it already, the Data Science Insider is a 100% free newsletter that the SuperDataScience team creates and sends out every Friday. We pour over all of the news and identify the most important breakthroughs in the fields of data science, machine learning and artificial intelligence. The top five, simply five news items. The top five items are hand picked. The items that we’re confident will be most relevant to your personal and professional growth. Each of the five articles is summarized into a standardized, easy to read format and then packed gently into a single email. This means that you don’t have to go and read the whole article. You can read our summary and be up to speed on the latest and greatest data innovations in no time at all. 
Jon Krohn: 00:16:00
That said, if any items do particularly tickle your fancy, then you can click through and read the full article. This is what I do. I skim the Data Science Insider newsletter every week. Those items that are relevant to me, I read the summary in full. And if that signals to me that I should be digging into the full original piece, for example, to pour over figures, equations, code, or experimental methodology, I click through and dig deep. So if you’d like to get the best signal to noise ratio out there in data science, machine learning and AI news, subscribe to the Data Science Insider, which is completely free, no strings attached at www.superdatascience.com/DSI. That’s www.superdatascience.com/DSI, and now let’s return to our amazing episode. 
Jon Krohn: 00:16:51
So yeah, you’ describing the tools you were building at Yhat were solving this kind of problem. 
Austin Ogilvie: 00:17:01
Yeah. So what we wanted to do was to basically the moment- 
Jon Krohn: 00:17:05
Doing minus one on a bunch of language translations. 
Austin Ogilvie: 00:17:10
Precisely. No. We had data science team, which I mean, data science as a professional class has come so far in the past 10 years. It was really a unique thing in 2010, 2011 to have a robust data science team. We felt very fortunate to have that and to be able to work with PhDs that know all about support vector regression, like this was very novel stuff. It was extremely frustrating to know what was possible to achieve with respect to opening the book of looms to a wider audience. We wanted to reach more customers and it took months and months to get certain models into production. So that was super frustrating. As product people, we wanted to deliver these experiences that we knew were possible and it’s very reasonable for a software engineer, a product designer to not know all of the intricacies of machine learning models. 
Austin Ogilvie: 00:18:13
From an outsider’s perspective, if a software engineer and a data scientist are both writing code all day, but these are quite different fields and operations that come cheap and easy in R or in Python may not be nearly as accessible in another target language. That was what our flagship product at Yhat was. It was you import the Python or R library and deploy any arbitrary expression as a very low latency API, very robust API that then any developer in the world is going to be able to use. 
Jon Krohn: 00:18:49
Right, right, right. Yeah. Super cool. And so, yeah, so even early on attracted … you were in the most famous accelerator on the planet, Y Combinator. So probably a lot of listeners have heard of Y Combinator. So it’s a program based out of Silicon Valley that takes in, I don’t know how many now, what’s in a Y Combinator class? Dozens of companies? 
Austin Ogilvie: 00:19:17
It’s quite big now. So they do two batches per year. Ours, we did it in 2015, several years after we’d started Yhat actually. And I think there was like 115 companies in our batch. 
Jon Krohn: 00:19:31
That sounds big, but I’m guessing [crosstalk 00:19:34]. 
Austin Ogilvie: 00:19:34
They’re even bigger now. It’s crazy. But yeah, it was a incredible experience. Yeah. I mean, Y Combinator is a very interesting model. They accept a certain number of companies into each batch and you spend three months out in Silicon Valley, although during the pandemic, they’ve now transitioned this to Zoom. 
Jon Krohn: 00:19:56
Oh, man. 
Austin Ogilvie: 00:19:56
And it’s just a laser focus period of 90 days where you’re working very hard. 
Jon Krohn: 00:20:02
Oh, it’s just 90 days? 
Austin Ogilvie: 00:20:03
Yeah. And we, at the end, we brought the whole team from New York. We brought everybody out for the last couple weeks, which was really fun. 
Jon Krohn: 00:20:13
Cool. And so, yeah, so they give you a little bit of funding, but it’s mostly about the mentorship and the network. 
Austin Ogilvie: 00:20:19
Yeah. So they give you 100,000 or so in investment and which is not that much, but it’s all about the network. And they provide you with very, very accomplished entrepreneurs who guide you through, based upon where your company is. Some are quite early stage. We were probably at, I don’t know, 30 or 40,000 in annual revenue. So we were among the earlier staged companies that were there and they really help you focus on what your particulars are that your business is going to need in order to take it to the next level.
Jon Krohn: 00:21:00
Yeah. In order to grow rapidly in, in order to blitz scale, probably in a lot of cases. 
Austin Ogilvie: 00:21:06
In a lot of cases. Yeah. 
Jon Krohn: 00:21:07
How are you going to find network effects, making sure that you have product market fit. 
Austin Ogilvie: 00:21:13
Yep. 
Jon Krohn: 00:21:14
And then you also, you probably meet if there’s even back then with a hundred other companies, you find other companies that you can work with. You’re solving different part parts of the same puzzle perhaps, or you can be a vendor to them or vice versa. I think a lot of that happens in these places, right? 
Austin Ogilvie: 00:21:33
Yeah. Totally. They encourage barters essentially among portfolio companies, but more than anything, you’re totally right. You find that most problems in business or technology are not unique. Every company has some unique problems that they need to solve, technical or otherwise, but for the most part people have encountered all kinds of patterns. And when you’re in a program like Y Combinator, it’s very useful to learn from the Y Combinator partners themselves, but also the other founders that you’re in the room with are overwhelmingly helpful in guiding you through various things that they’ve encountered that mirror a lot of your challenges. 
Jon Krohn: 00:22:19
Cool. Yeah. So that worked out for you guys. You definitely accelerated out of the accelerator. 
Austin Ogilvie: 00:22:27
We did, fortunately. 
Jon Krohn: 00:22:30
And then, yeah. So over the years, I mean, I guess I’m kind of skipping. I don’t know if there’s interesting stories that you want to tell along the way that would be interesting for our listeners, from Y Combinator to acquisition, but ultimately you were acquired by a big listed data company, Alteryx. So congratulations. That’s amazing. That was back in 2018. So yeah. So I don’t know if you want to talk about if there’s anything in particular in between there, part of the journey that was exciting. 
Austin Ogilvie: 00:23:03
Oh, it was quite the saga. I mean, in the very beginning, Greg and I had envisioned Yhat as being primarily a Heroku for predictive models product, where we imagined the data scientists would come sign up, put their credit card in, and be able to deploy their models to our cloud essentially. We quickly found that that business model wasn’t going to work for our customer base, that overwhelmingly companies that had invested in data science teams considered the byproducts of that investment to be critical intellectual property. They just weren’t comfortable deploying to our cloud. And also just like every other startup, the perceived or actual immaturity of the organization is viewed as a risk, especially for infrastructure and products like ours. That was probably the first year we struggled to make even a few sheckles. 
Jon Krohn: 00:24:07
You’re like, “Guys, we’ve got this super reliable. We can totally do this.” They’re like, “How long have you been doing it for and what are your clients?” You’re like ehh. 
Austin Ogilvie: 00:24:14
You got it. And then we literally took the advice of … one of our investors had been telling us every couple of months, you need to 10X your price and take down the SaaS model and go straight Enterprise. And that was very good advice that we didn’t listen to early enough, I would say. 
Jon Krohn: 00:24:36
That’s a good tip. 
Austin Ogilvie: 00:24:37
But the moment that we changed to a annual contract model, we did need to change the product to be able to deploy in any of the cloud environments that our customers were using, which was new technical territory for all of us at the time had never built anything that was meant to run across various versions of red hat and CentOS. But yeah, it was quite the saga. 
Jon Krohn: 00:25:06
Nice and yeah a happy ending. And so, yeah, I mean, it doesn’t get better than that to be your first company that you founded. And then five years later, you get acquired by a reputed company, big listed company, data company in the space. So that’s huge. And one of the things that I think is interesting about your story in particular is that you were a hands on founder, so you were technical. I knew that from our very first meeting where you were showing me, you were running Python code, R code in front of me to demonstrate how your products worked. 
Jon Krohn: 00:25:45
For me, I think that that’s really cool. I think that for me personally, I have a strong preference for working with CEOs like you, who are technical founders. I think that technical founders like that have a much more realistic understanding of the challenges that people face as they develop software products or in your case data science products wrapped into software in particular. 
Austin Ogilvie: 00:26:15
Yeah. I totally agree. Obviously there are limits. Not everybody needs to be PhD level familiar with every technical topic in order to start a business. But I think certainly in software land, it really does help in the early phase to appreciate what’s on the menu for possibilities when it comes to building something. It just helps you set realistic expectations with yourself and with customers, outside stakeholders, your investors, et cetera, a lot faster and a lot more accurately. That’s definitely true. That being said, I appreciate the compliment though. I would self-describe more as a hacker than anything else. I studied Arabic and Spanish as an undergraduate, very much a self-taught programmer. A lot of this community, SuperDataScience, what you put up on YouTube, your Udemy courses, this type of stuff, Drew’s book, Matthew Russell’s book, I attribute all of my technical development to that ethos, and it is amazing what one can teach oneself with respect to this whole ecosystem. 
Jon Krohn: 00:27:39
Yeah. So tell us about that. That was a question I had for you for later, but it just came up organically. So we might as well get into it. So how did you go from doing an Arabic and Spanish degree to somebody who’s hacking together data science solutions, founding a software company. I guess we have a little piece of the puzzle there that there was this intermediate part where you were a product manager, but how did you even know that? How did you know that you wanted to go from Arabic and Spanish to product management? And where in that journey did you start picking up using YouTube videos and Udemy for teaching yourself code? 
Austin Ogilvie: 00:28:17
Yeah, well, definitely roundabout way to find my way into tech. I was certain I was going to work as a spy for a brief moment. 
Jon Krohn: 00:28:26
Really? 
Austin Ogilvie: 00:28:27
[crosstalk 00:28:27] undergraduate. 
Jon Krohn: 00:28:27
I guess that explains the major. 
Austin Ogilvie: 00:28:28
Yeah, no, just very interested in travel, very interested in politics and political theory as an undergraduate. I got into tech as a result of an internship that I did between two years of my undergraduate. I worked for a very small at the time startup in DC, just the three founders in a house on M Street. The company’s called EverFi. It’s still around. They’re quite big now. And I just caught the entrepreneurial bug. I didn’t know that this was a profession, that it was possible to start coming companies. That experience led me to seek others like it, which is how I found my first job out of college at On Deck, the tech-enabled lending company that I was talking about earlier. 
Austin Ogilvie: 00:29:20
And you just spend a few hours with a group of engineers building something, and at least for me, I found the whole thing to be very interesting. It was frustrating not to know how it all worked. 
Jon Krohn: 00:29:35
Right, right, right. 
Austin Ogilvie: 00:29:37
So that’s how I got into it. And the resources fortunately were there. We’re very privileged to live in the time that we do if you’re interested to learn about design, about product building, about engineering, about data science. 
Jon Krohn: 00:29:58
Almost anything.
 
Austin Ogilvie: 00:29:58
Almost anything. 
Jon Krohn: 00:29:59
It’s an unparalleled time in history. I find sometimes I’m out at a cocktail party or whatever, and those haven’t really been happening much in the pandemic. I end up in these situations where people will say things like, “Oh, you know, that’s something we need at a time like this. It’s never been this bad. It’s a good thing that at least this one person’s doing something good with the way that everything’s horrible in the world.” And it seems so crazy to me because if you look at the data, in almost any respect over the last few decades, certainly over the last few centuries in terms of literacy, in terms of democracy, in terms of lifespan, in terms of quality of life over a long lifespan, in terms of just being able to put food on the table, more and more people on our planet, a greater and greater proportion of people on our planet are able to eat every day and to live a long, healthy life. Even in that one respect, things are getting better all the time. As a specific, on a smaller scale, it’s not as big as being literate or staying alive, having food to eat, but access to information. I mean, I guess there’s a flip side where there’s so much information that I think a lot of people get caught up in garbage information. 
Austin Ogilvie: 00:31:31
Indeed. Yes. 
Jon Krohn: 00:31:33
As we experience in American politics [crosstalk 00:31:36]. 
Austin Ogilvie: 00:31:35
There’s such a thing is too much. 
Jon Krohn: 00:31:37
Yeah, there’s so much information that you can just end up in a whole other world and a whole other reality, which is a problem, but there are still a lot of resources, in terms of learning resources, for teaching yourself basically anything, but in particular data science stuff. 
Austin Ogilvie: 00:31:57
Totally. I’m very bullish on professional education platforms, Udemy, Coursera. Some of the most talented engineers that I’ve ever worked with have come out of boot camps, flat iron school and [inaudible 00:32:13] Center and so forth. I don’t think it’s always obvious to someone who doesn’t come from a formal engineering education to understand that it is accessible to jump into tech and to learn these skills. And insofar as your technical career takes you down a road where you need a master’s or a PhD, that is available down the road too. 
Jon Krohn: 00:32:43
Totally. Yeah, do it online. 
Austin Ogilvie: 00:32:45
I couldn’t recommend more just if you’re interested to learn to code, sink your teeth in early. Don’t be intimidated by it. 
Jon Krohn: 00:32:54
Yeah, totally. It’s like, yeah, if you’re listening and you’re in the kind of situation that Austin was just prior to Yhat where you’re around engineers or data scientists, and you’re looking at their screens and you’re like, “Oh, that looks so complicated. I could never do it,” you can. 
Austin Ogilvie: 00:33:12
You definitely can. 
Jon Krohn: 00:33:12
Anybody can do it. Yeah. You just need to start somewhere and yeah, and you could ask that very engineer or data scientist what they recommend for that particular problem.
 
Austin Ogilvie: 00:33:25
Totally. 
Jon Krohn: 00:33:25
Yeah. What blog post should I check out? What book should I check out? What online curriculum should I check out and you’ll be on your way. Yeah. So very cool. So after the acquisition of Yhat by Alteryx, as is common when these acquisitions happen, a lot of the team and often particularly the leadership team stay on at the acquiring company. And so you stayed at Alteryx for a year. I don’t know if there’s anything interesting in particular you’d like to share about that. 
Austin Ogilvie: 00:34:01
Yeah. So we hadn’t planned, it was on our objective to sell the company. We had been at it for four and a half years or so. And in 2017, there were a number of Alteryx data scientists that had signed up for free trials. And they were all working on a hackathon internally. I had gotten some airtime with them and had a blast. One of them socialized that they were working on their hackathon project using some of our technology and it made its way up the food chain. And so Alteryx is very interesting company. They predominantly, their flagship product is a desktop app, drag and drop data manipulation. There’s some machine learning tools that are in there but imagine if Panda’s had a UI. What would that look like? 
Jon Krohn: 00:34:55
A click and point UI, yeah. And it’s a no code or low code tool. Right? 
Austin Ogilvie: 00:34:58
Exactly. And very, very popular, very a large user base. They had been studying the data science landscape, which was emerging in the years proceeding. And basically they had come to the conclusion that while they have this amazing desktop app that serves for many use cases, one class of data professional, that there is this emerging class of other data professionals that don’t work exclusively with drag and drop tools or with tools with UIs. And that there was this whole class that was underserved by Enterprise software to support the Open Source tools like Pandas and Python and various R ecosystem tools. And they were actively looking for a way to advance their product roadmap to serve that population, which is why they looked at us. They probably looked at, at the time, there was a company called Sense. I don’t if you remember [crosstalk 00:36:04] founded around the same time. DataQ was founded around then, same with Domino, sort of DataRobot. It was a bit of a different kind of thing. But these data science platforms were very attractive targets for all tricks to add quickly, add tools that would support different use cases that they needed to support for this particular code-friendly type solutions. 
Jon Krohn: 00:36:36
Nice. That’s a really good explanation. What was that like? What was that like to have your small knit team, many of you having grown up in the same apartment, working on top of each other? You probably would’ve been involved with hiring almost everybody at Yhat. 
Austin Ogilvie: 00:36:57
Yeah. 
Jon Krohn: 00:36:58
And then all of a sudden you’re part of a much larger organization. Was that easy? 
Austin Ogilvie: 00:37:04
It was an exciting journey and a challenging one, obviously. So we were 17 people at the time of the acquisition. So we weren’t a very big team. Yeah. So all Alteryx was actively looking to add to their portfolio of products, tools that would serve data scientists that are working with code instead of just the drag and drop audience. We were an attractive target. They loved our roadmap and what we’d been planning to build. Precisely what they were looking to bring into their own product stack. And yeah, we decided to go the MNA route, spent a year putting the team in the technology, into the fight, trained all of the sellers on the acquired products. And it was an amazing experience. I couldn’t have used all of the venture capital dollars that we had raised previously as an independent company to buy my way into some of these sales opportunities. Doing a million dollar deal with Nike to equip them with our software was an incredible experience. 
Jon Krohn: 00:38:13
Cool. And then after a year at Alteryx, you did something that sounds really nice to me as somebody who, I don’t think I’ve had a vacation in nine years, so you took some very well deserved time off to travel. 
Austin Ogilvie: 00:38:29
Yeah. So after I stayed for a year, I took six or nine months off, traveled all around, went to Uzbekistan, Ukraine and Lake Como and Nashville and yeah. 
Jon Krohn: 00:38:45
Nashville. 
Austin Ogilvie: 00:38:45
Nashville. 
Jon Krohn: 00:38:46
What a weird place to pick. 
Austin Ogilvie: 00:38:48
Big bluegrass fan. 
Jon Krohn: 00:38:50
Yeah. So how did you, I mean, you picked places like you went to Chernobyl in the Ukraine, right? 
Austin Ogilvie: 00:38:56
I did, yes. Yeah. That was a dark day. Yeah. I just picked places that have always been interesting to me in particular Uzbekistan has always been a bucket list trip. 
Jon Krohn: 00:39:14
Yeah. It’s not a country you hear a lot about. 
Austin Ogilvie: 00:39:15
No, it’s not, but I’m very interested in Central Asia. Yeah. Very interested in the Mongols in history. Yeah. 
Jon Krohn: 00:39:25
Cool. And then after taking that time, or maybe even during that time traveling, getting some head space, started to percolate, I guess, on the next thing, which emerged in 2019 you founded Laika. You’re founder and CEO of Laika now, and it’s going pretty well. So you just raised a $35 million Series B led by JP Morgan growth equity partners, bringing the total amount raised to $48 million. I know you’re doing something in compliance. 
Austin Ogilvie: 00:39:58
Yeah. 
Jon Krohn: 00:39:59
So tell us more about that. How did you stumble upon this opportunity? What are you guys doing? 
Austin Ogilvie: 00:40:04
Yeah, so definitely even while I was at Alteryx, I knew that I intended to build something else early stage. It was a matter of figuring out what I wanted to work on. And one of the problems that we encountered both before the MNA when I was running Yhat, as well as after the MNA as part of the bigger company, in going to market with a B2B software application, especially with an Enterprise sale is you encounter a lot of security due diligence and compliance scrutiny that for many early stage companies and even late stage or publicly traded sales teams is very, very challenging. There’s rigorous expectations that buyers have with respect to their software providers and Laika is designed to solve that problem. So we are a compliance automation platform. You basically hook Laika up to all of the services you use to run your business, your cloud provider, your ticketing system, your code repositories, et cetera, and Laika will aggregate automatically all of the things that are relevant to proving your compliance posture for the purposes of sales or in many cases, regulatory reasons as well. 
Jon Krohn: 00:41:31
I can see the huge value in this. There’s a huge opportunity. I mean, I can see, Austin, there is this huge problem. If you are a gigantic company like Microsoft, you can afford to have a dedicated team to handle these gigantic compliant requests that come through when somebody’s considering working with you. You could have an individual compliance request for an individual potential client, could have 100 or 200 questions. Some of these questions are easy to answer. Maybe take the person only a few minutes, but other ones take hours to answer. And you need to talk to lots of technical experts to dig down and figure out do we have scenario A or scenario B? So yeah, so if you have a small company, which I reckon is your target market primarily if you’re a smaller company, then you don’t have this army of people that can deal with these kinds of requests. 
Austin Ogilvie: 00:42:25
Yeah, yeah, yeah, no, you’re exactly right. The average software company is not going to be familiar with the breadth of security and compliance requirements and expectations across every geography, every industry. What is expected from a buyer at Yale New Haven is quite different from an Australian bank, which is quite different still from a California based software company. And how is it possible for the average SaaS business to stay abreast of all of these requirements and to add insult to injury, it is overwhelmingly the case that the person most likely to be asked questions about encryption standards and information security policies is a salesperson who is the least likely to know about these concepts. So, yeah, there’s a lot of friction and a lot of challenge bound up with these due diligence processes, these procurement processes. 
Jon Krohn: 00:43:24
Yeah. And so in terms of the ROI, the return on investment of working with a company like Laika, is that it means that more deals can go through, right? 
Austin Ogilvie: 00:43:35
Totally. That’s what excites me so much about this. It’s not obvious to most entrepreneurs. It certainly was not obvious to me when I was running Yhat, that there is a relationship between compliance and top line revenue. 
Jon Krohn: 00:43:52
Totally. 
Austin Ogilvie: 00:43:54
Of course compliance is a protect the bottom line function that’s quite obvious. Everybody paying attention can notice that, but there really is a growth lever to be found in building a robust compliance program within your organization. It helps expedite deals. It helps build trust with partners and customers. It helps you enter new markets faster. And so in that sense, it is a growth machine. 
Jon Krohn: 00:44:25
Yeah. And I think that’s even, I hadn’t actually pieced that together in your signature. So maybe it’s your slogan for the company. It’s like the founder’s unexpected growth lever or something like that. Right? 
Austin Ogilvie: 00:44:37
The growth lever most founders overlook. 
Jon Krohn: 00:44:39
Right. That’s the one. Yeah. That’s a really good point. It’s like there’s this opportunity by working with a company like yours to smooth out compliance requests. You can get through them more quickly. You can assure prospective clients more wholeheartedly, more convincingly, and more deals will go through that otherwise might have slipped through the cracks. 
Austin Ogilvie: 00:45:02
Yeah. It’s also a good idea. The tech industry overall has some maturing to do when it comes to not playing fast and loose with the rules. Obviously I’m an entrepreneur. So there’s a balance between doing things sluggishly and slow and doing things recklessly. And we help our customers find that right balance in a stage appropriate way. Help them grow in a mature way, and it doesn’t over complicate or break the bank either. 
Jon Krohn: 00:45:37
Right. In a way it actually, it can allow the leadership at a smaller software company to feel more comfortable with being reckless, because you know that you’re within the bounds of the constraints that a company like Laika is providing. 
Austin Ogilvie: 00:45:54
Exactly. That’s the peace of mind that I would’ve wanted at Yhat, and that hopefully we provide to our customers as well. 
Jon Krohn: 00:46:05
Cool. So given all of the data science work that you were doing in the past with Yhat, and then later Alteryx, I suspect that there’s probably a data science angle here with Laika as well. 
Austin Ogilvie: 00:46:16
Yeah, there definitely is. So a lot of the data that we’re ingesting from our customers’ day to day operations represents a very unique and novel data artifact. Nobody has previously compiled all of this compliance relevant data in the way in which we have. There are lots of interesting things we can do with this. One of the things that we’re doing that’s really cool is processing arbitrary text questions that appear as part of an RFP or as part of a security compliance review, and giving our customers the ability to automatically respond accurately with verifiable compliance data pertinent to those questions. So if you’re a salesperson and you encounter a question about hypervisor security or encryption standards, you may not know about these things. And if you have Laika it doesn’t matter because Laika provides you a UI that automatically will answer such questions in a way that’s verifiable. 
Jon Krohn: 00:47:21
Nice. Yeah that sounds like a complicated, but also very cool and very powerful natural language processing technique, falling into the question and answering realm of the NLP field. 
Austin Ogilvie: 00:47:32
Indeed. Indeed. It’s going to be cool though. 
Jon Krohn: 00:47:35
Yeah. And hugely useful. So, you love talking about data science, you’ve done a lot of data science hands on in the past as a hacker. You still hacking away at Laika? 
Austin Ogilvie: 00:47:47
Always. Always. They don’t let me push code to production anymore. 
Jon Krohn: 00:47:52
Yeah. Nobody [crosstalk 00:47:53] that either? 
Austin Ogilvie: 00:47:53
No, always. I’m always interested to check out the latest and greatest. Python is a core part of my day to day, always find a reason to script something out, you know? 
Jon Krohn: 00:48:06
Nice. Yeah. Do you have any particular tools that you use that listeners might not be aware of that you recommend? 
Austin Ogilvie: 00:48:19
Snowflake is the best database that I’ve ever used. It’s amazing. It’s just wonderful tool. There’s a desktop IDE called PopSQL, Pop S-Q-L, very cute name. That’s probably my newest data-related tool that I’m using. 
Jon Krohn: 00:48:40
Very cool. I don’t think we’ve had, at least since I’ve been host of the show for the last year, those are two tools that’ve never come up before. So those are cool. So I guess in the PopSQL IDE, that’s specifically for working with structured querying languages. 
Austin Ogilvie: 00:48:55
Exactly. It’s a SQL ID that’s designed for analysis. You’re not given a UI to explore the table schema as much as you are given a UI that lets you plot things quickly. 
Jon Krohn: 00:49:11
Oh, cool. 
Austin Ogilvie: 00:49:11
Sensible default plots, that kind of stuff. 
Jon Krohn: 00:49:13
Yeah. I’ve never seen that before. I’ve never seen a querying tool. Yeah. That’s the kind of thing we’ve come to expect. Just like RStudio before you created Rodeo. So for R and Python, respectively, this is the same kind of thing for SQL. 
Austin Ogilvie: 00:49:29
Indeed. Same thing. 
Jon Krohn: 00:49:31
And actually, it’s amazing that I haven’t even thought to look for something like that before. It would have to exist and I’m glad it does because the only language in data science that’s more popular than R and Python, SQL. 
Austin Ogilvie: 00:49:43
SQL. 
Jon Krohn: 00:49:43
SQL. Yeah. Cool. Great to know about that. And I’m sure our listeners appreciate it too. Now you’ve talked about learning about data science, coding through online tools. And I know from a conversation that we were having just before we started recording that you’re really big right now into learning about games. So like the Unity gaming environment. Tell us about Unity and why you got into these things. 
Austin Ogilvie: 00:50:12
I just like games a lot, board games, video games. Just like the data science community has made available unbelievable resources to learn all kinds of things about different flavors of data science, same has happened in the evolution of game design. So the one I’ve been learning the most is Unity. There are other game engines out there that are equally accessible, but yeah, I’ve been having a lot of fun with it. 
Jon Krohn: 00:50:43
Unity is a famous game engine for deep reinforcement learning problems. Did you know that? 
Austin Ogilvie: 00:50:47
I didn’t know that. Tell me. 
Jon Krohn: 00:50:49
So deep reinforcement learning is, as you’re probably aware, and maybe many listeners are aware, it’s a absolutely fascinating area of data science. It’s the closest thing we have to an artificial intelligence system today that can make a complex sequence of decisions in a real world environment. So we use deep reinforcement learning algorithms, for example, to play board games. But they’re also adept at exploring environments and playing shooter games like, I don’t know, I grew up playing like [inaudible 00:51:27], but that kind of game. And so there are tools online. And in fact, in my textbook, Deep Learning Illustrated in chapter 13, which is on deep reinforcement learning, I talk about the main options available for you to deploy your deep reinforcement learning algorithm into. So there are environments, Open Source environments, that allow you to play relatively simple games or two dimensional games like Atari video games. 
Jon Krohn: 00:52:02
Once you get into the 3D realm, the Unity game engine is used disproportionately. It’s the standard for creating 3D environments for deep reinforcement learning algorithms to play with. And so things like gravity, things like grasping in three dimensions. Yeah. All this kind of stuff that you need to do in 3D, the Unity game engine facilitates that, and a really interest byproduct of that is that it allows you to go from having … So you can train a deep reinforcement learning algorithm how to behave in a simulated gaming environment provided by Unity. But then that can translate to real world capacity to navigate an environment or to solve a Rubik’s cube, to grasp an object. So, anyway. 
Austin Ogilvie: 00:52:55
Fascinating. Is it like a Unity out of the box set of reinforcement learning tools that are there? I’ll have to check it out. 
Jon Krohn: 00:53:04
Yeah, exactly. So I haven’t done it myself. It sounds like together we could probably … 
Austin Ogilvie: 00:53:10
Sounds like we have a weekend project here. 
Jon Krohn: 00:53:12
Yeah, exactly. 
Austin Ogilvie: 00:53:12
Perfect. 
Jon Krohn: 00:53:13
If you can provide the understanding of Unity, they are really good for listeners, as well as potentially you, two friends of mine, Laura Graesser and Wah Loon Keng, they wrote a book published by Addison Wesley called something like Hands on Deep Reinforcement Learning, or Fundamentals of Deep Reinforcement Learning, I should remember it better and we’ll get it right in the show notes because it’s a book that I reviewed. I was an editor for the deep learning components of it. Oh, and that’s a critical thing is that the thing that makes deep reinforcement learning, the reason why it has that name, the reinforcement learning part is related to a kind of problem that you can solve. Reinforcement learning problems are characterized by the actions that take impact the data you get back. So this is different in data science or in machine learning. More commonly we have supervised or unsupervised learning problems. And in either of those cases, your data set doesn’t change based on model outputs. But with the reinforcement learning paradigm, if you’re thinking about playing a video game or playing a board game, every output that the algorithm has changes the environment and the response of the opponent, or the way that the 3D environment looks. 
Jon Krohn: 00:54:41
So reinforcement learning has to deal with this continuous change to make decisions despite the continuously changing environment. In recent years, deep learning, so the use of artificial neural networks to solve these reinforcement learning problems has really taken off. And so we call that area deep reinforcement learning. And anyway, this book by Laura Graesser and Wah Loon Keng is brilliant as an introduction to the field of deep reinforcement learning. And I know they specifically built an Open Source tool that makes it very easy to integrate with Unity games and create deep reinforcement learning agents, or in fact actually use their pre-built deep reinforcement learning agent. Just like in scikit-learn where you say, “All right, I want a random forest, or I want a regression model,” You can say, “I want a deep reinforcement learning agent,” and you can read their book to see what kinds of types might be useful for a particular game. And then just use their pre-coded deep reinforcement learning agent that they put together in PyTorch and apply that to whatever Unity game that you’d like to. 
Austin Ogilvie: 00:55:53
That’s very cool. 
Jon Krohn: 00:55:56
Anyway, so your show, and I just did a lot of talking. 
Austin Ogilvie: 00:56:00
It’s all good. 
Jon Krohn: 00:56:01
Hopefully interesting. So nice. Yeah. So that’s cool. So yeah, over the years, I don’t know if you’d ever call yourself a data scientist, maybe particularly, but you’re definitely a technical founder of a data science company, definitely a hacker, and so you probably have an interesting insight into what makes a great data scientist. 
Austin Ogilvie: 00:56:31
So for me, what makes a really great data scientist is someone who’s product minded. Frankly, I think the relationship between product management and data science is often overlooked by many people. These machine learning techniques are born out of academic research. Many of these techniques were first theorized in the 50s and have been around in principle for a long time. And what makes a data scientist, in my view, a data scientist, is that this person is not in the academy performing research for research’s sake, but instead for applying those skills to the real world somehow, and I think we are all accustomed to these Netflix and Amazon’s recommendations, Spotify. Name your flavor of exciting or elegant user experience. This is the world we’ve become accustomed to and data scientists have created that world. And yeah, so when I look for in a data scientist is somebody who sees those opportunities and has a particular ability to articulate to normal people the value of certain types of very complex work that normal people are not going to understand, that they do don’t need to understand, and helping businesses build and understand those use cases and build towards them. 
Jon Krohn: 00:58:13
Yeah. Hugely important in practice. If you trained as a data scientist in a formal program, so a formal education program program like a bachelor’s degree, a master’s degree in data science, a PhD in machine learning, or you self-taught on Udemy or Coursera or whatever, and you haven’t been exposed to commercial problems, it’s very easy to overlook this, which is, as you say, I mean, you brought it up, is something that makes a great data scientist, and it could be the most important thing. It actually in practice for the most part, it doesn’t matter that you know the most sophisticated approaches. It doesn’t matter that you know exactly how the linear algebra or the calculus works. So that can be useful in lot of applications. And it can expand what you can do in terms of creativity with a solution potentially. But the most important thing is how can I deliver value as quickly as possible to this client or to this business problem in my own company? It’s that delivering value, this kind of product mindedness, and I guess another part of it that you’re saying there is also being mindful of exactly what the user experience will be. 
Austin Ogilvie: 00:59:31
Yeah. Or at least encouraging the other stakeholders that are involved in building something, encouraging them to understand the capabilities of your work and the limitations. That’s when the real magic happens is when a product designer who doesn’t understand any of the intricacies of data science work is working successfully to build and deliver those kinds of magical experiences that we all love. 
Jon Krohn: 01:00:00
Yeah. A good contrived example that I’m sure happens all the time is you could, as a data scientist say, “Okay, well, you’ve given me this difficult, natural language problem, this difficult machine vision problem or something.” And so you use the latest and greatest, you go to Archive, you find the most fancy object detection model possible, or the most elaborate natural language processing model. And it gives you 1% better accuracy than a regression model or something. 
Austin Ogilvie: 01:00:34
Totally. Knowing when the juice is not worth the squeeze and to be able to identify the low hanging fruit. Data science work is very expensive. This is not cheap R&D work. Being able to iterate quickly, get something live, prove it out. If it works, improve it later, that kind of attitude towards the work is also contributes to my view of what a great data science team is all about. 
Jon Krohn: 01:01:05
Yeah. And that is all 100% spot on, but where I was actually going with that is that even that choice to use the very, this enormous- 
Austin Ogilvie: 01:01:14
Leading edge. 
Jon Krohn: 01:01:15
Yeah. It might mean that in terms of user experience is terrible because maybe with the regression model that was 1% less accurate, it took a thousandth of the time or a millionth of the time to get you the result. So as a user, you get this onscreen immediate experience. Whereas if they use the big, deep, latest cutting edge model, it’s like they’re waiting around 10 seconds to get the result or something. 
Austin Ogilvie: 01:01:40
Yes. 
Jon Krohn: 01:01:42
Yeah. So I love that answer. Thanks for that. So we talked earlier about how you transitioned from an Arabic and Spanish degree into being a technical founder. There’s one question that I have left related to that, which is how are you so successful so young? Do you have tips and tricks for listeners? You came from a non-technical background, very quickly self-taught how to hack enough to understand these problems. And then just dove feet first, founded a company, was CEO of the company. After five years, you’re acquired by a big listed, well-respected data company, and now you’re onto your second startup and you’ve already raised almost $50 million. So how could a listener who wants to get to that end point of building successful technology companies, maybe a data science company, and whether they already have data science skills, or maybe they don’t have technical skills at all, how can they bridge that gap over the next few years? 
Austin Ogilvie: 01:02:57
Well, I’ll happily accept, humbly accept these compliments though I don’t know how deserving I may be. I would say that just asking lots of questions, being really generous with your time and being willing to just be very honest with asking for help is a big part of it. In the early days, for me, it was a lot of generosity of those on my team, spending time explaining concepts to me, as far as adopting new technical skills. With respect to entrepreneurship, just know that you can do it. Definitely get a job at a startup if you’re interested to start your own company. Work somewhere early stage. The earlier stage, the better, if what your goal is, is to start a company. 
Austin Ogilvie: 01:03:55
There’s no better way to learn how it all works than just by doing it. And a lot of people don’t know that startup jobs are even available because they may not be listed on the jobs page, but let me give you an insider’s view. Startups don’t have the resources to send representatives to career fairs. If I get an email from somebody directly, every single time I take the meeting, I don’t care what the role is, every single time. I’m not special at all. If you find a company that you’re interested to work for, email the top person in whatever the field is. You want to work in engineering, email the CTO. That’s my advice. And I think that you will have very high response rates. Because startups need people that are really ambitious and really hungry and curious. We don’t have endless room to hire an infinitely large team. We only really have room for people who are very, very excited to be there. 
Jon Krohn: 01:05:07
Yeah. I suspect a part of having that cold reach out be successful, I agree with you 100% that it can be a successful approach, but probably a key that was implicit in what you’re saying, but maybe we could make explicit is that your email needs to be tailored to the company that you’re emailing.
 
Austin Ogilvie: 01:05:30
Yeah. Definitely you can’t just lob a generic email over the wall, but if what you’re interested in is working for a particular startup, just email directly and tell them. 
Jon Krohn: 01:05:44
Yeah. Cool, awesome tips. Ask lots of questions. Be generous with your time. Be honest about asking for help, about needing help, and know that you can do it and work at an early startup. Those are great practical tips that anyone can execute on. All right, Austin, every episode I ask for a book recommendation. Do you have one for us? 
Austin Ogilvie: 01:06:09
I always love, I probably read it once a year, Eastern Approaches by Fitzroy Maclean. It’s a memoir. It’s one of the people who Ian Fleming based James Bond’s character after. 
Jon Krohn: 01:06:24
Oh, no kidding. 
Austin Ogilvie: 01:06:25
Very interesting. Adventure. A lot of adventures. 
Jon Krohn: 01:06:28
There’s a real person that James Bond is based off of. 
Austin Ogilvie: 01:06:30
Yeah, yeah. 
Jon Krohn: 01:06:31
No way. 
Austin Ogilvie: 01:06:32
Yeah, Fitzroy Maclean. 
Jon Krohn: 01:06:33
Wow. 
Austin Ogilvie: 01:06:34
Yeah. Eastern Approaches. 
Jon Krohn: 01:06:35
Cool. There you go. 
Austin Ogilvie: 01:06:36
Love it. 
Jon Krohn: 01:06:37
All right. And then clearly you are a fount of valuable knowledge on data science, on tech startups. How can people follow you? How can listeners keep up with what you’re doing? 
Austin Ogilvie: 01:06:50
I’m on LinkedIn and Twitter. 
Jon Krohn: 01:06:51
Nice. 
Austin Ogilvie: 01:06:52
Twitter, Austin Ogilvie. Same on LinkedIn. 
Jon Krohn: 01:06:55
Yeah. We’ll be sure to have those links in the show notes. All right, Austin. Thank you so much for being on the program. It’s been wonderful and hopefully we can catch up with you again sometime in the future. 
Austin Ogilvie: 01:07:07
Thank you. Appreciate being here. 
Jon Krohn: 01:07:15
Well, I really appreciate Austin making the journey over the Brooklyn Bridge into Manhattan to film this episode with me in person. It made for a strong personal connection while filming that I hope was tangible to you while listening to the recording. In the episode, Austin described his journey into and successfully out of his famed Yhat data science startup, including practical tips how switching from a consumer focus to an Enterprise focus could potentially enable you to 10X the price of your software, dramatically increasing revenue. He also talked about how the prestigious Y Combinator program accelerates startups capacity to find product market fit and network effects, as well as to collaborate with other ambitious startups. He talked about how structuring your compliance such as with the tools provided by his firm, Laika, is the growth lever, enabling companies to expedite and increase the success rate of software deals. 
Jon Krohn: 01:08:10
And he provided his tips for being a successful technical founder, whether you come from a formal technical background or not, namely be generous with your time, be honest about needing help and ask for it. Know that you can do it and work at an early stage tech startup to gain exposure. As always, you can get all the show notes, including the transcript for this episode, the video recording, any materials mentioned on the show, the URLs for Austin’s LinkedIn and Twitter profiles, as well as my own social media profiles at www.www.superdatascience.com/535. That’s www.superdatascience.com/535. 
Jon Krohn: 01:08:49
If you enjoyed this episode, I’d greatly appreciate it if you left a review on your favorite podcasting app or on the SuperDataScience YouTube channel. This episode is particularly interesting because I am in person with a guest, which has only happened for the second time. So you might want to check that YouTube recording out. And yeah, I encourage you to let me know your thoughts on this episode directly by adding me on LinkedIn or Twitter, and then tagging me in a post about it. Your feedback is invaluable for helping us shape future episodes of the show. All right, thanks to Ivana, Mario, Jaime, JP and Kirill on the SuperDataScience team for managing and producing another inspiring episode for us today. Keep on rocking it out there, folks, and I’m looking forward to enjoying another round of the SuperDataScience podcast with you very soon. 
Show All

Share on

Related Podcasts