Originally aired on August 18, 2021 @ 1:30 PM - 2:00 PM EDT
Yes We Can is a recurring series presented by Cloudflare Co-founder, President, and COO Michelle Zatlyn, featuring interviews with women entrepreneurs and tech leaders who clearly debunk the myth that there are no women in tech.
Dora Jambor is a graduate student in Computer Science at McGill University, and the Quebec AI Institute (Mila), working under Joelle Pineau and Dzmitry Bahdanau. Her research is in semantic parsing, the task of automatically translating natural language to executable programs. Specifically, her research is in developing AI models for semantic parsing that are more data efficient, and can better generalize to novel scenarios they have not encountered before.
Currently, Dora is also a Visiting Researcher at ServiceNow where she works with Dzmitry Bahdanau.
Previously, she worked as a Scientist in Residence at the Creative Destruction Labs and NextAI where she advised various startups. Prior to that, she was a machine learning engineer at Shopify working with recommender systems for personalization and search problems.
To watch more episodes of
Yes We Can — and submit suggestions for future guests — visit cloudflare.com/yeswecan
Women in Tech
Hi, everyone. Welcome back to this week's episode of Yes We Can. We've taken a couple weeks off this summer and I'm feeling refreshed. So I'm super excited to be here with Dora Jambor. Hi, Dora. Welcome. Welcome to Yes We Can. Hi, thank you for having me. I'm very excited to be here. Yeah, well, I'm super excited. Today we're going to talk about artificial intelligence, we're going to talk about data science, we're going to talk about entrepreneurship and all the things that you're super excited about. We're going to talk about the world. You're based in Montreal right now. And so really excited for you to be here. So thanks for making the time. Yeah, me too. Thank you. Good. Well, let's just jump right in. And so just a reminder, for those tuning in, if you have questions as a way to submit questions online, just follow the prompts. Or if you have suggestions for guests that you'd love to see on Yes We Can, you can email yeswecan at Cloudflare.tv. And let's jump in with Dora. So Dora, today, right now, you're currently doing your master's in computer science, as well as the AI program at Mila, which is an AI Institute in Quebec in Canada, in beautiful Montreal, I went to undergrad there. So I love that city. Can you tell us a little bit more about your focus at graduate school? Yes, of course. So yeah, I'm coming to the end of my master's degree. And I'm potentially starting a PhD in the coming months. And so basically, I'm affiliated with McGill. And because of that affiliation, I'm also at the Mila AI Institute, which was founded by Yoshua Bengio, who's a very well known figure in AI and deep learning. And so basically, I'm working under two fantastic supervisors. One is Joelle Pino, who is the head of all of AI research at Facebook. She's a superwoman, you should definitely have her on your podcast. And then the other co supervisor is Dima Badnau, who is a very well known person in deep learning for some of his foundational work in the field. And my research focus specifically is on semantic parsing, which is a sub branch of natural language processing, where basically the focus is on translating natural language, questions and sentences into structured meaning representations. And what I mean by meaning representations is basically programs, you can think about programming languages like Python and SQL queries. So overall, like broadly, you can think about my field is like, we're trying to figure out how to automatically translate human language to code. I love this, you use that. So you did a really good job describing very technical things in ways you're like, wait, I kind of understood what you said. So thank you, Dora. You use it. So I just want to go back to this idea that you're focused on semantic parsing, because I think that this is a term that I've heard before affiliated with artificial intelligence, you start to talk a little bit about what it is. But maybe we can just like break it down even more. So let's just spend one more minute, because I really want the audience to be like, I want them to leave feeling like, oh, I just learned more about AI. Because I think it's something whether you are a business leader earlier in your career, middle of your career, late in your career, it's an important AI is becoming more important to all of our jobs. So go back to semantic parsing. So it's natural as a part of natural language processing, where you kind of take looking to something into code. So can you give us some examples and maybe how that might be used in the real world just to illustrate this? Yeah, for sure. So basically, like the reason why I ended up focusing my research on this was very much driven by a real world problem from my past. So basically, before graduate school, I was a data scientist at Shopify. And as part of my job, I very often had to work with product managers. And product managers are the people who are amazing at, you know, coming out with the right business questions and making decisions that very often require insights from data sets and from the databases that we have available at the company. And what happens is that the process of like the product manager accessing these insights from the databases is very indirect, because often the product managers are not fluent in these programming languages. So for them to be able to answer a question or for them to be able to make a data driven decision, they need to go and find a data scientist who will then has to pause all other tasks and write the SQL program or whatever program to retrieve the necessary information from the databases. So it's a very indirect cycle. And it causes a lot of delay in terms of like answering these questions and decisions. So semantic parsing is trying to automate this aspect of the data scientist job. And that's kind of like, I'm trying to do this thing such that I can automate what I used to do. And yeah, I think it's an extremely commercial application. And I think a lot of companies are looking at this space for that reason. That's great. That's great. Sorry, my kids just ran in and they want to know if they can watch TV while they're getting their hair cut. The answer is yes, you can. Okay, there we go. Okay, this is live, as you can see, anyhow. So I was trying to do multitask there. But anyhow, but what you're saying, Dora, about how when you were product managers trying to get information or business owners, we have this where it's, hey, there's some data and I need to understand what segment of our customers are doing this. And they do have to go to our data science team to be able to make the SQL query and want to get it. And it takes time, because those people are doing other things as well. And so you got to file a JIRA ticket, and it gets into the queue. And maybe it's done later that afternoon, best case scenario, but more likely it's happening next week. And so how does semantic parsing and AI, how does it solve that problem? I mean, ideally, like what I think could be a future sort of product around this is just have a platform where the product manager can write a question and interact with the data directly, such that they can actually come up with maybe even other questions and just, yeah, have the process of writing the code, take that out of the equation and just make this interaction more direct. I love that. Well, I think that that translation and just making it just work like magic would empower a lot of people to build even better products for their customers or get questions, answers, and would really speed everything up. And I think whether you work in a business, you can think about a lot of use cases where you're like, well, that would be much better than having to file another JIRA ticket. So good. Thank you for explaining to us what semantic parsing is, very important. As you've explored artificial intelligence and computer science in your grad schools, you might be pursuing a PhD. I mean, what elements of the field really interest you? I would love for you to share that with the audience. So I guess like just keeping it quite high level, like I don't want to go into like very specific technical details of all the things I have looked at through my AI career. But I would say like one of my visions or one of my interests that has been present for a long time is how do we build more natural interfaces for people to interact with technology? So you can think about like even simple things like a mouse and a keyboard. Like those are like barriers for someone to be able to like access a computer. And I think there is a lot of like information and creativity being lost by making humans like use these unnatural interfaces. And I think partly why I'm studying and doing research in natural language processing is because I think language is something that is that comes very naturally to humans. And if you're able to use language to interact with the computer, I think that that can tap into a lot of creativity that right now is maybe not possible. Hmm. You start to think of a lot of sci-fi movies and books that I've read of like interface becomes gone and just kind of gets embedded in us. Then for sure. Yeah. Start to think we'll go to dreaming in a little bit. Yeah. But I mean, the world is going towards that direction. If you think about like even things like Alexa and Google Home, like people don't really want to do mundane things and like use their eyesight even to look at information. Like it would be so much easier to just, you know, you're driving a car, maybe not while you're driving, but say like you're walking on the street and you want to just know about your emails or know about things that you have going on for that day. It would be so nice to just have like use language to access that information. There is a lot of idle time where you kind of say, can I use that idle time to do other things while I'm doing it? And I've had that where I'm walking to get a coffee. You don't want to look at your phone, but is there another way to get the information of things that are coming up? Or can I take care of something on my task list or even while I'm driving a car I'm driving, but I'm like, instead of listening to the radio, I could listen to my calendar read out for the afternoon. It is, it is, it feels like we're on the cusp of, of a lot of those tools and programs getting a lot better. If we can get the tech, crack the, the, the, some of the, the, the, the technology side, because I think there's still some problems to solve. For sure. Yeah. Yeah. Yeah. It's good. So you've, you know, obviously you're really passionate about artificial intelligence and thinking about all these kind of where that field is going. How did you, maybe you have a really interesting story about how you ended up doing your computer science graduate degree and potentially your PhD and studying at Mila. So maybe you can share with the audience. I'd love for you to share the very windy path that you've taken to get here. Cause I think it's very inspiring. Yeah. I guess my background is extremely untraditional. I I've yet to find other people who have had a similar path. But yeah, basically when I was a teenager, I was very much thinking that I don't really want to follow what everybody around me is doing. So I decided to study abroad and to leave right after my high school graduation, which was something back then that nobody really was doing. And I went to Amsterdam in the Netherlands and I started a program in economics which was then the only English program in at the university of Amsterdam and all other programs were in Dutch. So I started my program and very early on, I realized that actually I was more interested in the math and statistics and maybe computer science but I didn't have a way to switch my degree because they were not available in English. So then in my final semester, my university started a collaboration with Harvard and they actually offered one of the, one of the computer science courses at my university. So I could take the course from Harvard remotely and then get credits at my school for it. So I started that and that was super, super fun. And then after finishing my degree in economics, I just completely changed and I spent six months just building up my technical skills and studying like every day, kind of as if it was my full-time job and building up my kind of technical portfolio. And then I got lucky to have to get accepted to this program in New York city. It's called the Rikers Center, which is, they call themselves like a writer's retreat for programmers. And it's basically this like unstructured space, completely self -directed. Few people get accepted, like really, really smart people. And the idea is that when you put a lot of smart people in one space, then really cool projects will come out of all those collaborations. And there's just one shared goal that the goal of this program is to make you a better programmer and you can interpret that in any way you want. So this was back in 2016. And this was also when deep learning like started showing really cool results and people started thinking like, okay, I think AI is gonna like change a lot of things in the tech industry. So I basically like devoted my time in this school to learn machine learning and deep learning and just build up my technical portfolio focusing on machine learning. And then right after this program, I got a job at Shopify. So I moved to Canada and I worked there as a data scientist focusing on recommender systems. And throughout like two to three years, I was always thinking that I want to go back to grad school and to really like sharpen up my skills in computer science. But then in 2019, I was just like, okay, I'll submit my application to one school and see what it says, if I get accepted. And then Joelle Pino, who is a fantastic supervisor and person, she took me in, she took me under her wings. So I left my job and then I started grad school. Wow, I love this. I love, I mean, it's just, I feel like one of the themes I've heard on Yes, We Can is this idea of I just started to fall in love with it, or it was really fun. And I kind of kept learning more and more. And I just, I mean, think about it. You didn't say this, but you grew up in Hungary. So you went from Hungary to the Netherlands that led you to New York City. And then you said, oh, Shopify, I mean, an amazing company. We had the woman that runs data science and engineering for Amazon, who I know, you know, from when your time there, Ottawa, Shopify, now you're doing your undergrad and have a great mentor. And it's just how one, that one class, that one collaboration where somebody between Harvard and the in the Netherland university, like decided to set that university to Amsterdam, decided to say, hey, we'll offer this. Like that one reciprocal course, I mean, changed your life in so many ways. Yeah. It makes a huge difference to have access to that actually. It would have been a very different path if I don't take that course back then. Really. And it's just, you know, I think about that as all of us who like in your day job, like somebody at Harvard and the university Amsterdam created that together, you know, they worked on that. And sometimes I think, do they realize how, how, how, how they've impacted, you know, again, your life in the world because of that, that project. And I think about that all the time at my day job about these things add up little things out up to big things over time. You did a really good job illustrating that. I think it's nice to give people the feedback. Like I remember a few years after my graduation and I actually emailed one of the guys who made this partnership happen. And I just told him like, Hey, heads up. Like I'm so happy. Like I'm doing this thing. Thanks to you guys. So it's nice to give them the feedback, I guess. I'm sure you probably be in his day. So that's so nice and good on you for doing that. You know, one of the things, and I actually don't know the answer to this. So I don't mean to put you on the spot, but just, you know, it's interesting because when you said after high school, I went to do my studies in a different country and that was very uncommon from where you grew up. And I had a similar experience where I grew up, where most people went to school locally. And the fact that I was going actually to Montreal to go to school, different province across the country, people thought, do they have a program you can't do here? They, people didn't understand that. Do you ever, do you ever think back, think about why you, why you had that thirst to kind of go try something different when everyone around you was doing something, they had a different path set out. Do you ever think back to like why you saw the world in a different way? Hmm, that's a big question. We ask the question here on Yes We Can. Yeah. I, I guess I always just seeking out adventures and just, I wanted to do things differently. So kind of like I've had this like inner rebel in me throughout my life, like not just as a teenager, but I even now I find that I'm, I always feel a bit different from other people around me. And I think we'll talk about like entrepreneurship, like that's another topic where it's like the risky, the adventurous choice. And I'm much more like leaning towards that rather than like the, the comfortable and like the more safe route. So I think it's really just like a personality trait that some people want things to be different. Yeah. Or go pursue, see what else the world is out there in the world. It turns out the world is a big place. There's lots of opportunity. It's good. Yeah. Yeah. Yeah. Okay. So just coming back to you know, so you're now you're at grad school, you're there, you're associated with Mila, this amazing AI Institute. Can you tell us just for the audience, actually, I don't even know the answer. Like how, how has your, how is the semester, your week, your day, how, how has your day structured as a grad student doing computer science and Mila? Is it how much time do you spend in class? How much time do you spend in labs? How much do you spend by yourself working on your ideas with your advisors? Can you just give us the audience a sense of how your graduate program has looked? Yeah. So I guess like my experience has been very untraditional because of the pandemic. So yeah, through the pandemic, it was a lot of time spent alone. But in normal times, like you'll have a lot of reading groups and seminars and presentations and just like in-person interactions with other students, especially because we have this really nice space at Mila where you get to see all the other students. But yeah, so basically like you have to take some courses and while you take courses, you're still expected to do your research. And of course, like you'll spend less time on that. But once you finish up your courses, you'll just devote all your time to research. And normally like you meet your supervisors once a week and do like a little research update. And if you have collaborators, then you might meet with your collaborators more often through the week. For me personally, I've mainly worked with my supervisors. So it's quite like individualistic. So most of the time I'm going to be working on my reading papers, writing code and yeah. And basically like you can choose to interact with the community or you can choose to not interact with the community. It's kind of like pretty unstructured. So it's totally up to you. Self-directed. There's a lot of self-directed. So what's your favorite part? My favorite part is actually like the in-person meetings where we brainstorm and have a whiteboard and just really like flesh out ideas and solve like technical problems together. I really enjoy that aspect. That's great. That's great. You know, when you're describing, it brought me back to grad school. I did my business MBA and we started working on Cloudflare as a school project. That was, I guess it wasn't, I didn't think it was a, it was a business plan project, but kind of our research project. And that turned into Cloudflare and now that's my job and becomes my company. And you just don't know where these projects sometimes go. They can turn into really big things. Kind of what we were talking about before. Yeah, for sure. If you're in the right room with the right people, like you don't know what will come out of it. Yeah. Sometimes one of those whiteboarding sessions or conversations lead to really big things. And that's what, what happened to me. So I understand that. So, you know, you mentioned entrepreneurship and I know we got really excited as we were prepping for this session because not only are you in computer science and you're an AI, you're falling in love, but you're also really passionate about entrepreneurship. And you talk about becoming an entrepreneur after grad school. And so maybe you can start by telling us some of the sectors that you're interested in. Yeah. So there are really like two, two like areas. It kind of vague on my mind, like which direction I'm going to take. But one is like the semantic parsing space, right? So the very natural choice there would be to build some kind of platform for product managers or for people that need to interact with data and deploy some kind of like semantic parsing model there. So that's one area. And then another very unrelated kind of tangential area that personally is very exciting to me or like something I'm passionate about is the space of wellbeing, mental and physical wellbeing. And they're like the main question I'm interested in is how do we make like wellbeing more accessible to wider crowds and how do we make it such that say an individual doesn't have the resources to pay for a therapist or to work through trauma or I don't know, get emotional support, but still needs help. Like how do we use AI technologies to help those challenges? So, yeah, that's another place that I'm thinking about. Well, I think both are, both need innovation, both need better solutions. I think there's opportunities on both sides. And so I'm excited to see, we are all excited to follow along and see where Dora goes post grad school, but I think both are really interesting and I can already see your tagline, empower everyone to be a superstar at work for the first one, semantic parsing and then helping all of us be the best versions of ourselves. I feel like that's a big emerging trend that all of us want to be better versions of ourselves. So I love both your passion. And so this idea is pursuing entrepreneurship postgraduate studies like at McGill and Mila, is that common or are you also kind of paving and charting a new path there? I wish it was more common. Unfortunately, I think like what I see, and I don't know if this is true for other AI labs, but I think there are a few big corporations, say like Google, Facebook, Microsoft, Amazon, and so on. And they run these big research labs and it has become very natural for graduate students to wrap up their degrees and just like join one of these companies. And I think that's very dangerous in a way, because now you have all this amazing talent coming out of these universities. And this AI talent is actually going to be owned by a very select few companies. And so I've been involved with a few like incubators and accelerator programs. And I see that there is this, there's so many ideas and there's so many like founders like seeking AI talent and so many problems are waiting to be solved. And this talent is not there because they go to these big companies. And I wish that there was a better way to send our amazing talent from Mila to these smaller companies and to have more collaboration in that space. And I think like if you're a Mila student, like it is very much possible because you have this Mila brand name with your name and a lot of incubators and accelerators trust Mila and they want to work with students from Mila. So for example, like through the last years, like I was involved in doing like technical mentorship with startups in an incubator program. And I was also involved in like selecting and interviewing companies for some of these accelerator programs. And through these experiences, like you get a lot of exposure to like, what are the ideas? What are people thinking about? And where does my unique like skills that fit in? So yeah, it's being talked about a lot at Mila. And I think there is like effort going towards changing the ecosystem. But I think there's still a lot of work to be done. Yeah, it sounds like it. Well, I think they're lucky to have you because I feel like you're going to help pave this other or help make this, you want multiple paths when I think for the students coming out of Mila. And so it'd be great for entrepreneurship and joining earlier stage growth companies to be one of the paths. I think that would be a great, and I hope five years from now we're talking to her and you're like, it's changed. And I was part of the reason why it helped change because- I hope so. I'm willing to bet on you. Thank you. That makes sense. So you're learning a ton about computer science and AI and all these applications, which is amazing. It's incredible being used at large technology companies. And you've said, so how do you keep your entrepreneur skills and networks sharpened? So you talked about being associated with accelerators and incubators. Can you tell a little bit more? Do you read? Are there certain resources you read online that you can share with the audience? So very concretely, there are two programs in Montreal. One is called Next AI and the other one is called Creative Destruction Lab. I think they have other locations as well. So they are looking to work with Mila students constantly. So it's a very possible route for people to take to get exposure to the local ecosystem. And then there's also like AI for Social Good summer school, which is specifically like undergraduate students, women trying to build projects and products using AI. So I was also a mentor there. That's also a possibility for someone who's in this AI ecosystem. And yeah, I think just like we have a Slack channel at Mila, and there is a lot of information there. People do post like what companies are coming out and what are the trends. So it's good to just like stay on top of that and read once in a while. That's good. But just a reminder of the being curious and the rate of what you learn is a huge asset early in your career. And then just doing things that you want to do. Getting involved, showing up to those. It opens doors, you learn things, you leave expanded as a person. I think that's just a good reminder to do things. And I think it's also OK to not like stay on top of every information related to AI because it's such a big field. I think it's also nice to just if you're interested in a specific problem or you have some passion, like for example, for me, this like well-being is like very personal. Then I think just getting comfortable with the idea that it's possible to start something on my own and it's possible to use my skills for my own company instead of thinking like, oh, no, I must take a job and I must go to this big organization. It's good. It's good. OK, we have about 90 seconds left, a minute and a half, and I have to ask the AI expert because it's 15 years from now, if we can imagine a world, how will AI changed? How will AI have improved or changed our everyday lives? Yeah, I don't know if climate change is not going to destroy us by then, but if it doesn't, if we're still around, I mean, I my hope is like there will be a lot of innovation and drug discoveries and like disease prevention and disease treatments in 15 years. I do think that self-driving cars will be a solved problem by then and then more so in my space. I do think that my programming as a skill set will, of course, still be around, but I think maybe more people will have access to technology through these like text to code interfaces. So, yeah, so it's just like more accessibility to technology, I think, is what I'm expecting. More embedded, just just works more magic, just it just kind of works. There's less. Yeah, just kind of all around us, all encompassing. Well, this is great, Dora. This was amazing. I mean, it's hard not to feel so inspired after chatting with you. You've done so much. We're all excited to see what you do next. If we want to follow along where your Twitter, what's your Twitter quickly? It's Dora Jambler. Perfect. Yeah, at Dora Jambler. Awesome. Thank you so much. Thanks, everyone, for tuning in. My mind is spinning. We'll see you next week and thanks so much, Dora. This was so fun. Thank you very much. Bye, everyone.