Karen Sun is CTO of two companies – Dwelo, which automates apartment management, and Selesy, developing an online K-12 platform for every student in Colombia. She’s the rare person who can do it, with a background that includes classical music at Julliard, data science at MIT, and high-speed trading on Wall Street. A don’t-miss interview.
Christina Noren: I'm really excited by the guest we have today, Karen Sun. She's one of the people that I've encountered along my journey through technology that I respect the most, I enjoy talking and working with the most and have had the privilege of working together, and she's always working on amazing things.
For me she's the epitome of what we're after with software agents, which is people who are using their agency with software to do amazing things to change the world. So with that, Karen, tell us a little bit about yourself and what you're doing right now and then let's get a little bit into your background of how you got here.
Karen Sun: Yeah. Thank you so much for that kind intro, Christina. We have a lot of history together and I don't think we would have time to get into that, but that's been very much a part of my journey to get to today. I see myself as somebody who really loves to leverage technology to build public utilities. So right now I am the CTO of two companies. One is in the States and one is in Medellín, Colombia, which is currently my home.
In the States, I work at a company called Dwelo, which is an IoT company that builds smart community technology. So what that means is we're trying to help manage living spaces, kind of like apartments, outfit their spaces with smart, remotely-controllable devices, and this is to improve accessibility and comfort for tenants and also to kind of increase management and cost efficiency for property managers. So you can think of this as like energy savings, time savings. We're really, really starting to edge into how can we work together with a lot of the different utility and energy outfits in order to help the environment, so I just think that for the future it's inevitable that we're going to have these devices deployed everywhere, so we might as well be very cognizant about it and build a topology that people are going to have to live with every day that's secure and ethical and also actually improves the environment.
Christina Noren: And so it's going to be interested to get in to how that – and we'll do it a little bit later, but I can only imagine how managing, you know, large apartment buildings and all their systems is changing with the distancing requirements and threats of multiunit living, so that's going to be a really interesting one to get into in a minute. So what's your second thing you're doing right now?
Karen Sun: Yeah, you're absolutely right about all of those callouts. So here in Medellín I'm working with a venture called Selesy. Selesy is building a high-quality digital education experience for K-12 across the country, and this is to provide an alternative to typical schooling because the country anticipates that most of the schools are actually going to be shut down for at least an entire year due to COVID.
Christina Noren: It sounds like a very realistic projection.
Karen Sun: Yeah, definitely. It's really been interesting living here but then following the news in the States because Colombia is definitely approaching everything differently, a little bit more unilaterally. Of course there's a difference between the different stratas, so like if you're in a richer area versus a poorer area I think people kind of adhere to the rules more stringently if they're more well off, but the prevailing theme is that we are trying to be really strict with movement and trying to coordinate it all at once. And so I think a school shutdown here is going to be taken a lot more seriously than in the States, but that means you have a ton of children that are unengaged for most of the day.
So just a little bit more about it, because it's kind of unique, it is a public/private partnership between a nonprofit sector of the government in this department and two existing rather radical schools. One is called Holes and one is called Fontán. Right now we're focused on addressing something called the Derechos Básicos de Aprendizaje, which is kind of like the core elements of learning that licensed schools are required to address, and we are digitizing a storytelling pedagogy that's actually existed in Medellín for over 40 years, and we're going to be delivering that and testing it even starting a month from now, and I think this is only possible because we're in this really unique time where people and institutions are coming together super quickly to address large problems and they're doing it in a way that I don't think would've been possible before, like during steady state.
Christina Noren: Yeah. I mean one of the viewpoints that we've definitely had on this thing, that we've seen as a thread through all of this, is that it's a whole – you know, there's horrible suffering happening and going to happen, but just like World War II some things that wouldn't have happened for decades that are good are starting to take shape that will benefit us for decades to come.
Karen Sun: Yeah.
Christina Noren: So Karen, before we get into details of both of those – and this is our first sort of, you know, doubleheader on one of these – let's get to know you a little bit better. So, you know, I can say we met when – ironically it's a PR person that Paul, my cohost, introduced me to, introduced me to you, and said, "This has to be your data scientist for Aura," the startup we were doing together.
Karen Sun: [Laughs]
Christina Noren: You know, and you'd met at a _____ tech startup.
Karen Sun: Yeah.
Christina Noren: And I know you went to MIT where Paul was as well. So tell me your early journey and your interests and how you got here and catch me up to the present day.
Karen Sun: I don't know exactly how far back to go, because as you know it's a bit of a meandering path.
Christina Noren: We like meandering paths on this, so go for it and Paul can trim a little bit for time.
Karen Sun: So I always feel super fortunate and like my life is the most unlikely thing in the world because I don't think it was very directional. As you know, Christina, I spent most of my life studying classical music, so starting from I think when I was 10 I attended Julliard and I was studying violin performance, and that was pretty much most of my life. That was definitely my concentration. I don't think I was really super devoted to a lot of other kinds of academics. And then I did end up going to MIT for university.
I felt like that was a bit of a fluke. I actually really didn't want to go. I really wanted to study biology. I really wanted to do life science research and I just never saw MIT as a place that would really nurture that, but that goes to show how ignorant I was about universities in general and what they offer when I was a teenager. So when I got there I have to admit I was very influenced by just the people that I was surrounded by.
It was very entrepreneurial. I felt like half of the people I ran into knew how to code, and this was all rather foreign to me, but over time I became a little bit disillusioned with life sciences, which is what I was studying, and I did always love math. I have to admit I was really jealous of the empowerment that came with understanding technology and understanding how to code, so I eventually switched and I studied math and computer science, and then I kind of fell into a series of industries kind of just because they were in front of me. So I worked in pharmaceutics for a while followed by defense, followed by high frequency trading, and then really it was when I was working in high frequency trading I was feeling like if I projected my life out from that point on I was going to be pigeonholed, like both it would not be good for my own edification but also I wouldn't be learning the skills that could actually really make a difference in the world.
I was seeing a lot of brain – I felt like a lot of brain drain; just some of the smartest people I ever knew working on pushing money around. So yeah, around that point I decided to leave and just get more serious about learning software and learning how to build things end to end, and at this time I was also incredibly obsessed with doing things involving media because I just love kind of books and arts and music and a lot of my life was centered around trying to help other people develop appreciation for the entire universe of what's out there. And then pretty much within a year I was working at my first startup, Bookish, which was a publishing joint venture, and that was a great learning experience. At the same time I also got involved starting an ed-tech company.
At this point we were I feel like a little starry-eyed and naïve thinking that technology and websites can solve everything in the world, so at this time I very much believed in self-learning, being able to just kind of acquire any knowledge using the internet for free, so I was working with some friends on building an aggregator for free learning content on the internet and then kind of from then on I feel like my life kind of dual-tracked. So I was always working on learning tools for people, trying to figure out how to crack this nut of using the internet to find free learning materials, to up-level yourself in any way you wanted, and then at the same time I was kind of learning a lot about machine learning, NLP, building search engines, and I was either consulting for ed-tech or kind of building prototypes or working at a startup that involved media, and ultimately I ended up in San Francisco I think, like, four or five years after that and I was working with you on Aura.
Christina Noren: So for our listeners, at Aura Karen built the backend for a system that was taking memories of art experiences that our users had around the world and stitching them together and discovering interesting things and building a web of that knowledge, and it was just amazing work. We didn't have a lot of money and didn't have Karen for a long time, but we got to work together. [Laughs] Anyway, after Aura I know you wanted to travel a lot and do startups along the way, and so what's been the journey since then?
Karen Sun: Some of my fondest memories are sitting in the _____ coding, like in front of art that we were testing on. I just feel like that was a very surreal year of my life where your office was sometimes museums. So anyway, so I started to get to know a tech company called Quizlet. Actually, over two-thirds of high schoolers in America use Quizlet in their classrooms, and this is just, you know, very consistent with all the kind of education learning tool works I had done before, and I ended up joining them and building a data science team to leverage the 10 to 15 years of very granular study data that they had kind of accumulated over all the years. And I was building adaptive learning experiences.
Actually, the mode called "learn" in the Quizlet tool is based on that work. At the same time I had continue to be involved in education nonprofits either in the States or in other parts of the world, so I did spend part of my year traveling and kind of working on those initiatives. Eventually I really wanted to take a step back and get more experience working with hardware. I think that a lot of the things I want to build in the future are going to involve hardware and understanding supply chain, understanding how to manage distributed fleets of physical systems.
I feel like technology has gone through different cycles and I feel like we've gone through this, you know, huge software cycle that started really 20 years ago, and now increasingly the next set of hard problems and the set of data to collect is not going to be people in front of screens. It's going to be ambient – wearables, buildings, shipping, manufacturing, movement, the entire smart car movement.
Christina Noren: Well, that makes a lot of sense in that context. With the social distancing and COVID-19, everything from factory floor to apartment buildings we have to have software controlling hardware for us so we don't have to touch the hardware. I mean that's what seems to be a pretty fundamental world shift, and so it seems pretty prescient if you wanted to go that direction. Is that about the point at which you went to Dwelo?
Karen Sun: Yeah. Exactly. That was about two-and-a-half years ago. I was definitely excited by that challenge of kind of working on the inevitable. I now think working on software is a little bit like cheating because for the most part we deploy on these highly-available, highly-redundant, always-reachable cloud platforms, and when you want to troubleshoot something or when you want to scale something you can usually throw money at the problem to scale it if you have the money or, you know, you can usually reach something and figure out what's going. In the worst-case scenario you can, like, hit the reset button. I mean I'm really simplifying it here, but software in general has been a lot more solved when it comes to deployment and management than hardware has.
Christina Noren: So talk to us about Dwelo. So you're CTO there and you run the entire product and engineering team. What's the software that you've developed as hardware and what does it do and what was it before the pandemic and then what's changed in what you're building for your customers since the pandemic?
Karen Sun: That's a really good question. I'll give you a brief overview of our stack because it's actually a very comprehensive tech stack, like all the kinds of technology people usually touch. So we developed a hub or a gateway, which is essentially a single-board computer with a bunch of radios on it that sits in every single apartment unit in this last mile communication with edge devices such as your lights, locks, thermostats, leak sensors, things like that. So we have, you know, hardware expertise in house and we're always iterating on that. In fact, we're about to launch our next generation hardware hub. We do not develop edge devices right now; we integrate with existing smart devices, and then the bulk of our development happens software and client side.
So residents can use an Android or an IOS app to control everything about their home and also do things like set up links with voice assistance, like Google Home or Alexa, and they can use this app or those voice skills to control their devices and schedule them. On the property management side, I feel like right now that's really where a lot of like the value at scale is, the value proposition at scale. Property managers can access a web interface that allows them to understand what are the temperature settings in all my vacant units, how can I maybe remotely let people in or out, how can I integrate this kind of apartment management-managed software with other software in the ecosystem of managing communities? So these are things like rent roll, resident portals, perimeter access.
Communities are super complex to run, as you can imagine, and trying to not blow up the staff required for that is also rather difficult. The other thing we do is actually have to develop an entire internal suite of tools because we're vertically integrated. We work with construction projects sometimes as they're breaking grown or we work with them at the blueprint stage to understand how to install their devices. So we need to be training up an installation force all across America that is able to walk into a unit, understand how to install devices, provision hardware and the software that goes on it and make sure everything is working and then come back and troubleshoot it.
Christina Noren: What have been requirements, changes, or new areas of focus with the pandemic? Because I can tell you we're looking at a much simpler system for our little eight-unit self-managed building there, and it's – they just do the output control. That's incredibly – and we're being driven by it because we don't want to touch buttons in the elevator anymore.
Karen Sun: Exactly. I'd say there are two things. So one is just decreasing contact, which is unsurprising, and also kind of decreasing the requirement for time in office and presence for the people who work in the community. I'd say the most – like maybe the most accessible example of this is something called self-touring. So typically when you want to show an apartment building or you want to go see it, you interact with the leasing agent, you pick up keys, you go to the unit and you're together, and a lot of it is about controlling access and scheduling and being there for security reasons. There definitely has already been a movement, but it's increasing, to try to provide self-touring wherever possible.
So this means you can remotely schedule, remotely access as a prospective tenant a model unit so that you don't have to go and do a lot of this physical exchange, and ideally you wouldn't even have to touch the lock or touch any of the devices when you go in. Even greater would be if it could self – like units could self-clean themselves in between showings, but these are kind of like the high traffic spaces with a lot of nuanced access criteria and more security risk that we're focusing on developing for. For instance, we developed our self-touring offering very quickly over the past couple of months and I think we're going – we're set to launch that later in Q3, but that's an example of just making a beeline for something that's become very important.
Christina Noren: That's interesting. So let's get into that a little bit and then we'll move over to talking about Selesy. So I think on Dwelo, tell us a little bit about your team and how they work and how they're organized or managing them from far, far away, and then talk to me a little bit about how you deliver software to control systems in the cloud, I assume, and to all your devices, like what does software delivery look like? You know, how frequently are you doing releases and how do you respond to new requirements? Like how do you manage to launch a big feature like that in two months?
Karen Sun: I'll talk a little bit about the structure of the team first and how we work. It's something I'm rather proud of. I just have a really, really great team where every single individual takes a lot of ownership, which makes what I'm about to say possible. So under R&D we have product QA and engineering, and under product we have a lot of different hats, everything from product management to product marketing to UI/UX. Christina, I'm sure you're familiar with all of these hats.
And in engineering we don't have a huge team. Right now I think including everybody who's working we have 15 people, and just, you know, I described a tech stack earlier and there's so many components that we actually have close to a one-to-one or a two-to-one mapping of people to a specific area of expertise. So when it comes to engineers they really just have to own their code base, their area, and they really are the champion for kind of quality and security in that area. So whether it be data warehousing or the hardware design or the mobile apps, we really have like a leader for each one. And then we have a quality team which is very much an engineering team as well.
They have to figure out how to automate and virtualize the testing and the edge case scenario and the feature scope mapping of a staff that's very combinatorial. So, you know, we have a device catalogue of over 25 different things that might be installed and then we have actually I think six different clients that we're developing, and then all the different possible combinations of integrations in between those two layers, so you can't test everything every time and our quality engineering team just really tries to figure out how are we optimizing our testing pads every time we develop something new, and doing it very quickly and securely.
Christina Noren: So I assume you're using CI – obviously what CloudBees is going to ask you about – I assume you're using CI/CD tools and besides the devices things are deployed into the cloud. Are you doing continuous delivery?
Karen Sun: Yes.
Christina Noren: Are you using developer staging environments? Yes, yes, yes?
Karen Sun: Yeah. I can describe that really quickly because it's also rather interesting. I'd say we have three different – sorry, four different cycles of delivery that we really have to be cognizant of. So there's hardware, which is kind of a yearly cycle. [Laughs] It involves a lot of D&I and a lot of testing, a lot of rounds of testing before it actually goes out. We have the software that actually runs on the hub, and while that has an internal CI/CD pipeline for us to test internally, every single time we want to deploy software out to hardware we can't be sure of how it's connected, meaning we can't be sure if it's using ethernet or using cell, which is extremely expensive. So we have to be very cognizant of bandwidth when we deploy and we can't just deploy every release that goes out.
We actually have to kind of schedule them. So that happens usually at most on a monthly basis, but that's already a lot for people who kind of develop and have to update firmware and software and hardware. And then we have mobile apps, which most people are familiar with. You're kind of subject to the rules of the developer platform and usually that's like biweekly at most, I think monthly most commonly, but then everything software side, so whether it's our APIs or our web clients, that we deploy a couple times a week, sometimes a couple times in a day, you know, depending on what we're reacting to or what granularity a feature we're developing.
And for that all to happen we don't do continuous deployment because we always have a red button that the quality team kind of manages, and that's because everything we do is so sensitive from a hardware perspective. We need to make sure we've at least done some sort of smoke test against hardware before we actually deploy to production.
Christina Noren: I don't want to be standing outside my building and having my access code not work because you just deployed something.
Karen Sun: Exactly. That is the highest risk type of event for us, is like we lock somebody out and it's super cold outside – true story. Maybe your lights don't work. That hasn't happened to us, but we want to make sure that type of thing never happens.
Christina Noren: Do you do any feature flagging or does that just not make sense because of those factors?
Karen Sun: No, we definitely do feature flagging. So right now we mainly do it at the UI layer. I say we use it the most heavily with our installation fleet because we're always trying to develop new ways to auto-heal or auto-educate during the installation process or we're trying to figure out new, more efficient ways of installing and pairing devices, and then we want that to not be rolled out to a bunch of, let's say, third party installers right away who we haven't gotten a chance to talk to and we'd like to control who tries out the more experimental features first.
Christina Noren: So that's really interesting with Dwelo. So I have no idea how you manage all of that and manage to be CTO of a new ed-tech startup that is critical paths for learning in Colombia, so not to _____, but you're amazing. So talk to us about Selesy and what you're doing there.
Karen Sun: My involvement of CTO started there at the beginning of July. This is a project I've known about for a while, because as with something so large and a public/private partnership it's not like it sprung up overnight. It's been in discussion I think for two years and then a team was assembled four months ago. I do admit I don't sleep that much these days. After I joined, when we were looking at the school year and things became more urgent in the requirements – or rather the population we were launching to started to increase – I was like, wow, I have no margin in my days. I'm super, super fortunate.
My Dwelo team, as well as the Selesy executive team, are both completely aware of my commitments to the other projects and they were just really willing to work together to figure out a time split. I don't know if this would be possible without the urgency of COVID and without the kind of mission behind these projects, but either way I'm super grateful, and this just an example of something rather unlikely happening right now.
Christina Noren: There's a lot unlikely happening right now. So what's going on with Selesy? Tell us in more detail and then tell us how you're doing it.
Karen Sun: The goal is within a year to really provide a license, which means like a governmentally recognized school option for any K-12 learner in Colombia, and this has to be delivered digitally because of the nature of COVID, and already, you know, we're up against these questions of last mile delivery, who has internet, who has devices, but for now we have to start somewhere. The team assembled is half pedagogical, so people who've been very experienced teachers or have developed kind of school pedagogies from the ground up, as well as people with experience building technology, and the two directors right now, I and Andrea, the director of the whole project, we both have backgrounds in ed-tech. So she has a background directing the Holberton School, which is a pure learning pedagogy for developing programmers that was brought to Colombia a couple years ago, and I with Quizlet and some of my other past projects.
Right now we are taking something called the Fontán methodology. So what this is is something that students can independently navigate that teaches Spanish, mathematics, science, arts and history and geography, and it does so in the form of storytelling. So students sit down with stories that kind of try to immerse you in an experience and kind of a narrative, and along the way you are challenged to do exercises but also reflect on why you think answers to certain exercises are correct. So there's a lot of, like, self-reflection along the way, but it also was uniquely built for students to be able to do on their own with occasional feedback from teachers. This is something that we think is a good down payment on a system that students can access, you know, self-service across the country.
So we're trying to digitize this experience, which is always a little bit tricky. It's hard to take something that has been completely designed to be done in person and bring it online. In fact, I think you should never just try to do a direct translation but, you know, we are trying to create the same immersive experience digitally, and a lot of it right now on the technology side is figuring out, well, we have a very truncated timeline; what tools out there can we leverage? At the end of the day what's probably the hardest thing to do right now is manage hundreds, tens of thousands, millions of students in their learning journeys versus the actual implementation of the learning interaction.
Christina Noren: Can the platform do both? Is it intended to be the platform that is delivering this and watching results and interacting with school administrators and teachers and so forth?
Karen Sun: Exactly. The platform has to do everything, so basically it has to be at the point where like a parent, student, or teacher can just log in, create an account, so very self-service, and then through a series of diagnostics one of the things I'm excited about, because we are throwing away the age requirement, figure out where you need to start at least within this current map of content.
Christina Noren: Yeah. I mean that's interesting. I was talking to the CEO of an ed-tech company in the United States just a few weeks ago with maybe my semi-educated questions here, and we were talking about things like, you know, how do you make this successful to a kindergartner, figuring out what they're supposed to do next without having their parents have to be virtual teachers, and we were also talking about the problem of students transferring between schools and districts and classes and teachers and not having their results follow. It's just all sorts of interesting things. So I'm imagining you're dealing with all those kinds of feature questions.
Karen Sun: Yeah. You know, this is one of those cases where I'm super grateful to be seeing everything through two lenses. So first through the lens of my employees and the states, most of whom have many children at home right now, half of whom have decided they just have to homeschool their children over the course of the next year and are kind of in a privileged enough position to do so. And then here in Colombia just seeing just like a more unilateral approach to schools and whether they're open and what we should do. The hardest part is going to be figuring out how to engage parents and teachers, because at the end of the day if a child is very young they should not necessarily be the one to have to, I don't know, find all their options by themselves. They're going to need to be supported at home. And one thing I'm very proud of the team for is leading with community.
So the first thing we did when we stepped in the door was start to conduct a series of surveys, reaching out to parents and teachers and school administrators everywhere and just trying to understand what their reality is, what it would take for them to engage with the system, what they're looking for. Of course, it remains to be seen whether what people tell you in a survey is actually how they behave, but that is the first tack that we took when figuring out what to design.
Christina Noren: Yeah, it was an interesting challenge. I spent some time with School on Wheels, which serves homeless kids in California with tutoring and spent time – you know, I'm right by Skid Row and I was starting to tutor kindergartners and first graders at the Skid Row Learning Center and, you know, I'm not a _____. I haven't been in school for decades myself and I was so surprised by how much support the current curriculum and homework expects from an adult sitting with the child.
Karen Sun: Yeah.
Christina Noren: You know, these kids, they have committed enough parents to enroll them in the Skid Row Learning Center, and that's a level of commitment, but they're doing everything they can to survive and going home to a shelter or a tent, and those kids – you know, I haven't been able to do anything because I don't feel like I'm qualified to work online with kids who are sleeping on a cot in a shelter. It seems like such a hard problem.
Karen Sun: Yeah. Absolutely. I want to say that the team is super aligned on what we want to try to bring to children with this kind of unique opportunity, this unique crisis opportunity. So as you're probably familiar, schooling systems across most of the world have been rather factory set. There is this core set of content and this progression of knowledge you have to know, and most of it is memorization-based. What we're trying to do with this opportunity is bring more emphasis on inquiry-based learning, project creation, constructionist principles.
We're even kicking around these ideas of how do we make everything more pod-based so children have to work with each other on everything even if it's, you know, remotely? But there's this idea that we know you have to have a support system. This is literally going to be impossible unless every single learner actually feels like they either have a mentor or someone who cares about their progress to turn to. I know there have been a lot of AI plays in this space, but I just don't think we're there yet with that kind of technology. I think there has to be a human in the loop.
Christina Noren: Yeah, that's interesting, pods, but we can't explore that because we have to get into the software side of it, but this is just a fascinating topic. I can see why you're so compelled to do it.
Karen Sun: [Laughs]
Christina Noren: Tell us how you're organizing the software development and, you know, how you're going to make these crazy deadlines and what it's going to take to maintain the system once you launch it.
Karen Sun: Yep. Oh, wow. I wish you could help me answer that. [Laughs] So there are kind of two threads going on right now. One is digitizing this core system and then the other one is how to expand and really apply some more of these radical, principle-based constructionist learning ideas. So on the digitizing Fontán side, we need to immediately solve for three things; we need a platform that manages students in the self-service, allows both teachers, parents and students to interact; we need to be able to upload content and edit it; and we also need to allow learners to interact with the content. So those are kind of the three pillars.
The good thing is that people have been thinking about these problems for years and years. There are a lot of LMSs out there that really attack the first two problems, and then the latter, which is the interaction system, there's always this, like, ever-evolving ecosystem of people developing modular libraries so that you can kind of customize your own learning experience on top of. And I think right now because our timeline's so short we have to reuse. We can't build things from scratch. And that's what a lot of the initial investigation was, like what can we reuse and customize as quickly as possible and stay true to the type of learning experience we're trying to get?
Christina Noren: Are you adding people to the team and do you have a system, an internal instance to the system that you're updating regularly already and using?
Karen Sun: Yeah. Actually, I'll tell you this detail because it's a little bit funny. We have two internal instances right now of Moodle, which is like the trunk on which we're building everything. So one instance is localized to Spanish, and that's the one that's more externally facing and the whole team uses. But once you have a separate instance that's in English, because as with the world being so ethnocentrically siloed, all the documentation, most of the community, when you're trying to be really fast in development you just need the menus to be in English. It's hard to, like, figure out where all the analogous things are in Spanish. So we have these two instances running. We're mainly developing on the English instance and then porting over to the Spanish instance.
Christina Noren: Is there automation involved in that porting or is it very manual?
Christina Noren: Okay. Well we're definitely at time, and Paul, you've been quite through this because Karen and I get into the – switching antenna conversations together, but I know you're always really good at takeaways, so what's your – what are the things that are striking you from today's conversation?
Paul Boutin: I didn't have anything to say because there was just so much coming in. It's really impressive, Karen. What I hear is that being a software agent, which you definitely are, is a combination, it's actually a range that you bridge, the ability to code highspeed trading and the other creative aspect of Julliard and all that you've done. But then taking that into these situations where people don't think about software, don't think about math, and as you said, there's a lot of working with people aspects that we didn't go into in depth but we can see it, that the way to bring software to these people is to learn how to work their systems, understand their needs, and you hit something that really affects my work too, is the difference between people what they say when you give them a poll or a survey versus what they actually do when things happen for real. So that was terrific.
Karen Sun: Thank you. Thank you so much. I know that both you and Christina are also people of kind of mixed technical and artistic backgrounds and that this is very much a tribe, so I think earlier when we were talking about professional goals I'd say that's another one, is just understanding where to leverage software in places that other people might not see it.
Christina Noren: Yeah, and that's what this broadcast is all about. And it's funny, in coming up with this, we sort of had this idea that we're populating this Marvel universe of software superheroes kind of thing. [Laughs] And it kind of is that, and it was like almost everybody on this knows somebody else on this or, you know, their past lives, but I think there is a set of people that are not just software developers and not just entrepreneurs trying to find some software idea to make the best next buck, but they're really saying, "Okay, I've got this toolkit of my life experience, my interests inside and outside of software, my software skills and my mindset, and how do I go find a worthy problem and attack and – you know, understand it and attack it?" Karen, thanks a lot for talking with us today. This has been one of the really compelling ones and I'm excited to see what happens with both projects.
Karen Sun: Thank you Paul and Christina. It was my pleasure.