#32 Katharine Jarmul on being driven & focused on what you can do
⚠ The following transcript was automatically generated. ❤ Help us out, Submit a pull-request to correct potential mistakes
Tim Bourguignon 0:05 What is a good software developer? What do excellent developers do? There are probably as many answers to these questions as developers in the world. So let's ask veterans and newcomers what their story look like. Let's learn directly from them. Welcome to developer's journey. Hello and welcome to developer's journey, the podcast shine light on Developers Life from all over the world. Today we have a live because it again, we have Katrine jumbo here at the hepes compost conference in the UK. Hi, Terry.
Katharine Jarmul 0:45 Hi.
Tim Bourguignon 0:46 Um, you just gave a great keynote, the opening keynote of the conference in France have not quite full audio tutorial. It was too early for for
Katharine Jarmul 0:57 graphics.
Tim Bourguignon 0:58 Yes. Exactly a talk about ethics in the it. And I think we should come back to that at some point. But let's start with a bit about yourself. How did you come up to being in a position to speaking etc conference to be invited to hold a keynote at MIT conference? what's what's your background that led to that?
Tim Bourguignon 4:40 Um, since since you've been since you made this decision to go to devlopment. Yeah. How did this diversity of impact your decisions to where you want to work and I mean, was it was it clear from the very beginning that this is something you are going to seek out or did it slowly come into Your life and you realize we've had experiences that this is something for you. How did that come to be?
Katharine Jarmul 5:04 Yeah. So I mean, when I was a kid working on it, my mom actually also worked with the marketing side of a technology company. And my stepmother is mathematician by trade by training. And she also led an engineering team at Nortel, she worked in networking equipment. And so I had these strong female figures that you know, worked in tech that were good at math, most of my best math teachers were women. And so I was kind of in this bubble of Yeah, women are really great at math, I mean, obviously.
Katharine Jarmul 5:39 And,
Katharine Jarmul 5:40 and then when I entered university, this was, of course, a very rude awakening for me that I tended to skew the opposite way, both in my colleagues in terms of my peers, as well as the university staff. And I think that this, for me was the first moment that I realized that maybe math wasn't, you know, this kind of women led industry. And maybe computers weren't this kind of, like super welcoming environment, at least in the United States at that point in time, at least in the communities I was involved in. And I think this originally made me feel like okay, maybe there isn't a place for me here. Even though I really enjoy this work. Maybe this is not a nice place to be, and maybe I need to find somewhere else. And when I found my way back into it, I had an excellent mentor. His name's Ryan O'Neal. And I believe he runs a data science team now at grub hub, I think. But he really took me under his wing and spent evenings going over alar with me, and optimization equations, and so forth. So I could really get back into the math of the data science aspect, even though I had, you know, dropped that aspect, so to speak, from my major when I focused mainly on the economics and statistics. So as I wasn't as enveloped in the computer science optimization part of things.
Tim Bourguignon 7:04 So that was your time before the Western Washington
Katharine Jarmul 7:07 that was at the Washington Post. Yeah, he was. He was the team lead of the apps team, then of the state of journalism team. He set me up with my first Linux computer. And so yeah, it was good time. There's good time. And I think with that mentoring, it really made me feel welcome. And I helped found then a few years later, the very first pi ladies group, which is diversity in Python, and open source community, and I founded that with some of the women I knew, and we tended to be the only women at the meetup group said, complaining about it as usual. And then we said, okay, maybe we should do something about this. And we had a big dinner over at my house, and a cooked pig lasagna. And we met and said, Okay, what would could we do? And we said, you know, let's, let's form pi. Ladies, let's that sounds like a fun nameless forum. Hi, ladies. And let's put on some events and see what happens. And I'm happy to say that our second event was a hackathon all weekend, hackathon. And it was the first hackathon I've been to that had more women than men, we had 60% women, and I was like, where was everybody? Where were you hiding? And that was for us like a big game changer in terms of when you put that you support diversity out there. And when you put diverse folks speaking at the front of the room, you tend to just attract a more diverse audience. People feel welcome.
Tim Bourguignon 8:36 They do. You could you could you keep up the this this quote this person, quote, still, or I mean, in the pilot do suppose you if you do more and more Python oriented, oriented, science oriented? Can you still attract as many women?
Katharine Jarmul 8:52 Yeah, I mean, so I think one nice thing about data sciences, it has the maths aspect. And maps tends to be a little bit more diverse than some of the other STEM areas. And so I think one nice thing is I need quite a lot of female PhDs and Master's in mathematics, and also in bioinformatics and so forth. So I think that it has this aspect that is a little bit different than programming only, which I think still has a long way to go. Unfortunately, it depends on what country you're talking about. Right? So part of living in Berlin has meant I've met a lot of really amazing hacker in in and female programmers, particularly from Balkan states in the Eastern Europe states where it's more gender diverse in terms of what people study with computers.
Tim Bourguignon 9:50 That's cool. That's interesting. Um, I would like to come back to the mentoring topic. Just rolling back to our to your mentor Ryan. How did you find each other out or how did this mentoring relationship first originated? or How did that work out?
Katharine Jarmul 10:41 works. It's like, Okay, this is
Katharine Jarmul 10:42 better, less code is better. I remembered that much.
Katharine Jarmul 10:45 So.
Katharine Jarmul 10:47 And he was like, Who did this? And so he came and found me, and was like, did you build this? Did you build it? You know, yourself? And I was like, Yeah, yeah, I just put it together, because I thought it would be interesting with the story. And he was like, how would you feel if I set up a computer for you with some of the apps that we're building over on apps team? I said, Okay. Yeah, sure. I mean, I told him that, you know, at one point in time in my life, I studied computer science, and then I would be very open to helping out if it made sense. And he set me up with a boon to I believe, or maybe cu boon to at that point in time. And the whole stack of what they were using, which was at that point in time, actually the largest Django installation in the world, The Washington Post is running, which, for better or worse, maybe not the best place to start. And he gave me a commit key, and I took down production several times.
Katharine Jarmul 11:50 That was maybe not so great.
Katharine Jarmul 11:53 I learned quickly, it was just come running over. Oh, did you push something? Oops, sorry, rollback. So I learned to get pretty fast. And actually, we're using SV and at the time. But yeah, the essential basics of how to commit. And he really took me under his arm, I went over to just spend half time with the app team. And I learned from some folks on the team during pair programming and so forth. And he used to give me homework exercises. So building a linked list from scratch, all of you know, the classics, reading the what is it the art of programming, the ROB pike book, and so forth, I went through all the exercises there. And I started to really rebuild some of the basics that I remembered but make them more active in my in my daily life. And then I actually got recruited to lead my own team like that at a rival publication. And I went there and made more mistakes. I bricked my first server, there was some fun times. It was Red Hat Enterprise Linux for I believe, and I completely bricked it trying to install some newer Python tools, which is good talk. networks, they're fun. And since then, I've just been constantly learning and learning a lot on the job and learning from finding people that I think are doing admirable work, or who are really bright and trying to have them spend time with me. Now I seek mentors, and also tried to give back and mentor folks that I think maybe could use some advice or even just somebody to hear with them to listen to them.
Tim Bourguignon 13:37 How do you do that? Both seeking mentors and kind of, I wouldn't say seeking mentees, but nourishing in aquos mentee how do you do that?
Katharine Jarmul 13:48 Yeah. So I helped organize PI Data Berlin, which is a large conference on data science, in Berlin with open source tools that's run by a nonprofit organization. And essentially, the ideas that we can fund open source. So how can we give back to the community and help fund the open source development for quite a lot of the data science and machine learning tools that are in Python and in open source. And so that's the goal of PI data. And what I noticed when I first started getting involved, which has a little bit more than two years ago, that even though the community was diverse, the speakers were not as diverse. So didn't see as many speakers from underrepresented groups. And I thought, you know, perhaps this is a problem that I can help with. And so what was already there that I started actually working on is a mentoring program. And this mentoring program is not only for new attendees, so you're a fresh attendee, and you kind of want to know what is this whole thing about? How should I go about technical conferences, what should I do, learning about the hallway track and these other parts of conferencing This is an important part but also how Can we mentor new speakers? So how can we find new speakers that have, let's say, even fresh perspectives, or different perspectives? And how can we make them feel welcome to submit proposals and to, you know, speak at all of these conferences, and the mentoring program has really expanded. In the last year, we had more than 30 proposals come from the mentoring program, which is really great. We put on a few events, we put on a CFP hackathon. Basically a proposal fun, where we all get together in a big room, the PI ladies are there. We also have a hacker group in Berlin hacker in in called heart of code, they were helping promote, and we sit in a room and we just help people, it's basically open office hours, and we have some more experienced speakers there to help, you know, refine the proposal. Because I think you know, when you speak at a lot of conferences, you start to learn what people are looking for. And if you're completely new to it, this is uncharted territory. And then what we do is right before, we also have a present a THON, and you can come in, and it's a full weekend, and you can present your slides and get feedback, you can even just come in for a quiet place with coffee to work on your slides if you're leaving it to the last minute. Really not. And this is the weekend before the conference. And this gives you that extra boost, hopefully, of confidence of having some feedback. And it's also open to the community. So if you can't afford her, you didn't buy tickets in time to go to the conference, you can come get a sneak peek a few of the talks. And I think this has really helped create the community, we also have a scholarship fund that we use. So we have sponsors that fund scholarships for those that cannot afford their tickets and their travel. And this year, we were able to sponsor 20 attendees to attend who couldn't otherwise afford to attend. And so I think this, these are all just little pieces that I think make people feel welcome. And for me, this is really essential. Because if it wasn't for Ryan's help, and if it wasn't for a team that was so open and willing to have somebody break production occasionally, I wouldn't be here. And so the least I can do is give back a little bit.
Tim Bourguignon 17:20 My mind is just exploding with ideas, great ideas.
Katharine Jarmul 17:25 to try out three or four.
Tim Bourguignon 17:28 This is great. That's great. I'm asking this question, because I'm speaking a little bit later. And the first question I always get is, well, how do I find mentor? And my answer is usually by not speaking about networking. It would be like Like, like, like dating and talking about marriage right away?
Katharine Jarmul 17:47 Yeah.
Tim Bourguignon 17:50 And so I'm always interesting how people find the phone themself. And what I realize is mostly Well, there was one one person maybe with a bit of seniority, and they realize somebody was interested and just started nourishing the fire and say, Well, here's a piece. And if this person comes back, then we'll give him or her another piece. And we'll see how that goes. And this evolves in time into something. And so this is what you're confirming again. Yeah. It's really interesting. Yeah, yeah.
Katharine Jarmul 18:18 And, you know, it takes some patience. And I think it takes like a compromise on time from both sides. So now, a lot of times that people that I seek out, for example, I was recently in Singapore, and I stopped by the University of Singapore to this professor that I just completely admire, like in all of his work, his name is Professor Reza shokri. And he is working on quantitative privacy methods. So how can we quantify privacy? And how can we use adversarial learning lists his latest paper on determining how to create more private machine learning? And I'm just completely an office worth. And I just wrote him a little note, and I said, you know, I'm going to be in Singapore. I know, you're probably really busy. I admire your work. I loved this one paper. Do you have time? Do you have office hours, and I could drop by and we talked for nearly two hours, and it was fantastic. And for me, that was just so wonderful to be able to thank him for his contributions and his work. I hope for him, it was okay. But you know, you have to also as somebody looking for that mentorship, you have to also put yourself out there and understand that people are also busy and they have their own work and they have their own issues. So find that little compromise even if it's just a few hours. I try to make the most of it.
Tim Bourguignon 19:36 Yeah, I was thinking about what you're saying right now. I can't remember the last time somebody bothered me with questions.
Katharine Jarmul 19:46 Usually, usually people don't come to you. Yeah, the sensor in himself handled. Exactly. And people say oh, well, I know that you're too too busy. But I mean, throwing out a quick email or whatever it is, is the best way to communicate with the person and asking, Hey, do you have 20 minutes? I think as somebody, most of us have been mentored in some way. So I think there's a lot of understanding there.
Tim Bourguignon 20:10 And you just mentioned privacy. Maybe we should just go a bit in this direction. How did you come to being so important and social? So in interested in, in privacy, and now interested enough to make it your day job and, and create the startup around privacy? What's, what's the story behind this?
Katharine Jarmul 20:29 Yeah, so I mean, I think anybody that grew up with computers during, let's say, the late 90s, early 2000s, had a little bit of this allure of security, right security culture, and what a security mean, as well as a little bit of the experiences of the first internet, when it felt it maybe wasn't, but it felt slightly more anonymous, right. So I got my start in IRC channels, and so forth, I still hang out on freenode occasionally, and there was this idea that you could connect with other people that they might be also security and privacy minded or politically minded even. And that there was kind of this greater online community that was still aiming towards creating an internet where there was this free and open sharing, but that there was also. And on imaging, there was also the ability to potentially hide certain things about yourself or have some element of privacy. This was before, of course, the eve of the social networks, and the tracking of every browser across multiple parts of the internet. And maybe that was easier for people that weren't as young as I was to see coming. Maybe this was something that I just missed coming. But there was something that I feel like now is somewhat lost. And I miss the ability to feel as if I'm able to browse the internet in a more private manner, and that I'm able to interact with software and services, and choose, let's say, the level of information that I share or choose, for example, my perspective of the internet. And I'm a particularly privileged user, right? I know a lot about how technology works, I know about how tracking works. I earn enough money to have a comfortable life, I am white, I am educated. So the internet is not really that dangerous of a place. For me, it's pretty nice. Maybe occasionally you get charged more than somebody else. But that's probably like the most dangerous part. But when we see things like I mentioned in the keynote, Cambridge Analytica, and when we see people that maybe are less tech savvy, as well, as people that are far more at risk populations, we see how the lack of privacy can really create a dangerous place in internet for people, a place where they succumb to predatory advertising, a place where they're exposed propaganda. And obviously, these are really negative effects. And I'd rather, computers are awful sometimes, but I'd rather computers be a source of good I really like computers, I like working with computers, I want them to help make the world a slightly more positive place rather than an extremely negative light. And so for this reason, yeah, I think privacy really is a key point right now in terms of how we determine how we move forward in terms of computer engineering, computer science, and definitely data science. And I think by allowing people to choose privacy, and by allowing people the security of knowing that their data is also safe, that this is something that computers that we can do with computers, that would be a positive contribution. That would allow people to have a little bit more control over how they experienced the internet online and would hopefully deter folks from the types of propaganda machines like Cambridge analytic a story shows us.
Tim Bourguignon 24:17 I guess it's a loaded question, but how does hope look like in this in this future? Yeah, I mean, if I take my, my semester classes, yeah, I don't see hope in there. People tend to cut corners and not necessarily be evil by design. Yeah, be evil by mistake. And those corners are used and misused and will end and quite commonplace. And do you have a vision of that could be flipped on its head?
Katharine Jarmul 24:57 Well, of course, I do. I mean, I think everybody is going to come up with somewhat of their own conclusions on this. But obviously, I'm optimistic. And I must remain optimistic. This is just, this is my ami self, you know, this is optimism until the end. And part of my optimism is, for example, I recently founded a startup with my co founder and our vision, our goal really is to one could say, democratize privacy to make privacy easier. So how can we make privacy easier? And I agree with you, in general, I mean, the joke goes that programmers are lazy, right? We always find whatever shortcut to make it easier. So we can go back to whatever it was we were doing before we had to open our terminal and start working.
Katharine Jarmul 25:50 So maybe we open just
Katharine Jarmul 25:52 another terminal that we find more fun, but it's still a problem, right? And so rather than everybody in the world becoming privacy or security experts, how can we make this privacy by design or by default, right? And we all know, and especially those of you that interact with the security community, defaults are almost always set to completely open, completely free admin admin, let's all just have a great time. Because nothing bad will ever happen. If we connect an open database to the internet, it should be totally okay.
Katharine Jarmul 26:25 And
Katharine Jarmul 26:26 yeah, that's just, unfortunately, also the way that a lot of people are doing machine learning. So when you publish a model online, it has been proven time and time again, that extracting the information or making inferences about the training data, that essentially if you've connected in an open database to the Internet, and so I think what we're trying to build at KPI protect is tools that make this easy by default, how can we make things more private, that will still work for data science and machine learning. And so we've developed, for example, our first offering as pseudonymisation API that allows for structure preserving pseudonymisation. So how can we preserve hierarchies within the data or statistical relationships, while still privatizing the data in some ways. And so this is, for example, one of our first offerings, there's also a lot of other people working in this space. And this brings me hope I get a lot of exposure to this some privacy and security of machine learning space, this is a deep interest of mine. And I do think that there are people out there that are doing the research, they're doing the work on how we can build at least within the data science and machine learning community, more accountability, for security and more accountability for privacy. And I hope that this, you know, I think that this is inspiring for me. And I hope that this also inspires folks from numerous other fields within computers to do similar research and similar work.
Tim Bourguignon 28:00 So your first target is developers, or people in the big word, yeah, raise the awareness about all this and, and start thinking about it. And then we can talk about the end user.
Katharine Jarmul 28:11 Yeah, and try to make things simple and secure by default. Right. So whenever you're developing software, know that your user is probably not going to read the manual, and they're gonna use whatever's by default. And so if by default, it doesn't work until they set, let's say, a secure, encrypted password or this or that, whatever it is your service is providing, if by default, you make them take some action, this is maybe slightly better than just saying, okay, it's up to them. They got to read through the security section of this manual, which we all know, nobody's going to do. Maybe two people will do out of all of the people that use it. So
Tim Bourguignon 28:51 I was I was the, the, the resonance so far, are people reacting too long, to your talks and to your efforts.
Katharine Jarmul 28:59 Yeah, I mean, I think it's a mixed bag. And I think that's to be expected. So I think that it's really heartening, of course, I meet a lot of people who agree with me, and that's always nice. It's nice. Some people agree with you. And they want to talk about these issues. And they're excited to try them, and particularly those that are in industries where they already are thinking a lot about security and privacy. I find a lot of shared points of view, right. And this tends to be healthcare, banking, or finance and the security field. But outside of that, I think it's a hard sell. Because I think unfortunately, security, the onus has often been put on the developer to think of it and again, the developer is not usually a security expert. And so here you have this developer that's already probably doing five more things than they should be. That's already overworked or you have, let's say the IT person and they're already overworked and now they have to do pen testing too. And they have to think about All these all these other things as well. And I think when we overburden these teams with security is just one more thing on their plate or privacy is one more thing on their plate, we're doing a disservice to everyone, right? We're definitely doing a disservice to them. And at the end of the day, we're doing a disservice to whoever uses our services, because by default, we have privacy as a weekend project,
Katharine Jarmul 30:24 which doesn't work
Tim Bourguignon 30:26 as best. Yeah. So if it doesn't fall before the testing, we didn't have time for this, though, we will never have time for privacy and load testing.
Katharine Jarmul 30:39 Yeah, and I think, you know, this is hopefully something that from engineering management that people start to think about and really delineate a separate engineering team to something like this, as well as, hopefully solutions. Like what we're trying to build that make it somebody else's problem. Who only worries about that. Sometimes that's the best way to go is you don't want to worry about it. Well let somebody else worry about it.
Tim Bourguignon 31:05 The Korean market. I mean, people take care of it. Um, we're kind of reaching the end of the 10 books. I would love to come back to one aspect we didn't speak much about so you're now an intrapreneur. df employees.
Katharine Jarmul 31:22 Yeah. So right now we have only some part time, folks, we are looking to hire though and expand that will most likely happen in the beginning of next year.
Tim Bourguignon 31:30 Okay, so we've already hired, you mentioned you have been a teammate as well. And yes,
Katharine Jarmul 31:36 yeah, I've led a few teams. Yeah.
Tim Bourguignon 31:37 So I would like to do to come back to this idea. What is a good software developer? Or what kind of traits? Are you looking for? What are the people you you are searching for in the world? And you would like to work with them? And how do you find them?
Katharine Jarmul 31:54 Yeah, I would say, I look primarily for two different aspects. And they're difficult to find in the same person. So when I find them, I really hang on to those people. The first aspect is curiosity, and exploration. So want somebody that's curious to learn more, that's constantly questioning, maybe even questioning me, that's totally fine. I do things wrong all the time. So I want somebody that's curious, that kind of has this inquisitive nature that's wanting to keep learning and that's wanting to keep thinking. And even if I find somebody, let's say that has more junior skills, I know that that curiosity will eventually push them into senior skills, just almost by default, without even really having to do much is that curiosity is going to as long as you know, you put the path in the right direction that's going to really bring about a fantastic software engineer or developer one day, maybe even today, right? And so the Curiosity is something I always search for. And the other one is endurance, persistence, and endurance. And I this is why I think it's difficult for to find both because usually you're curious, but maybe you're a little bit impatient, I tend to fall on that side more by the endurance thing is something I've learned over time. And something that I often see in some of the people I admire the most in this field, which is the gunning, they're willing to see failure, and they're willing to push through that and try again, and they're willing to, you know, whatever, there may be perfectionist desires, and so far, they're willing to push through them, and keep trying and keep working and deal with these failures, our these shortcomings as they come and continue to just keep working. Right. And I think that this is something that you really need in a team is you need somebody there that's going to in the hard moments or in the everybody's there at midnight, because you have to ship and you're late, which hopefully isn't the case, right? If you're doing planning, right, but anyways, it sometimes is the case that there's going to be the person there that also helps keep the team focused that says, you know, we can do this. Let's keep going. We'll figure it out. And I think this is these qualities, I think are really admirable. And something that Yeah, I when I see them, I try to get that person as fast as possible.
Katharine Jarmul 34:17 Cool. Good to know.
Tim Bourguignon 34:20 I was about to ask you for advices for new new hires. But I guess that will be your advice. Do you have something another advice you would give to newcomers in our industry or people that already established but willing to willing to work?
Katharine Jarmul 34:33 Yeah, I mean, I think that curiosity and endurance cover it, but more specifically, I would say, try to stay driven and focused on what you can do, especially when you're younger developer even. So when you're older, and you're maybe trying to move into a different area or aspect of engineering or development. There's going to be things you don't know and there's always things that you don't know Nobody knows everything. And so really take that to heart and know that just because you don't know something doesn't mean you can't learn it, and doesn't make you a bad engineer, right? There's just always going to be more stuff that you don't know. And the more you learn, the more you learn that you don't know things. So I think just accepting and embracing imperfection and your inability to learn everything and do everything at once. I think this is going to be a way to stay in the industry for a longer period of time.
Tim Bourguignon 35:30 There I think I want to plug a book no the imposter and I think it's hi and Rob Connery. Hey, gorgeous book rump is isn't an established engineer. He is been working for decades in the industry. And he wrote a book going through the the basics of this Yes, degree that he'd never had, because he is self self taught. And he had this imposter syndrome his whole life about well, I didn't go to school through a CS degree. So I'm a bad engineer. And he now went back to the basics and learning this the hard way. But he and he describes a little story and why it's important, etc. It's exactly what you say. He's been a gorgeous engineer without us. And he's just now getting to it because he wants to, but he wouldn't need it. And I find this this really healthy, healthy way of doing it. Yeah. Great book. Cool. Thank you. That's, that's a very quotable thing. driven and focused on what you can do. That's a great, great service. I'm doing the stock. We forget to speak about something.
Katharine Jarmul 36:30 I don't think so.
Tim Bourguignon 36:33 Then there's a blessing. What's on your plate? And where can people find you?
Katharine Jarmul 36:37 Yeah, so right now at Kay, I protect her doing a few proof of concepts with some different industry partners, exploring how we can make privacy a little bit easier for folks a little bit more accessible. We will likely be hiring in early 2019. So if you're interested in working on privacy, even unquantifiable privacy, we plan on having a research department, so have your PhD within this space, please reach out if you're looking for postdoc opportunities. And yeah, you can keep up with me. I'm on an off on Twitter, sometimes I feel very engaged. And other times I must get away. But you can find me at K jam. And you can find our website and what we're working on at K I protect
Katharine Jarmul 37:25 comm
Tim Bourguignon 37:27 sounds like a great place to be to find you. And you'll be having talks as well.
Katharine Jarmul 37:33 Yes, yeah. Yeah. So I'll be at strange loop as well as Swiss cyber store Ma and go to Berlin. So if you find yourself at any of those, please feel free to come say hi. Or just if you're in Berlin and want to drop by the PI Data meetup you can always come say hi, the third week of the month is our meetups.
Tim Bourguignon 37:53 third week of the month. Great. Thank you so much. Thank you so much. So fun. And thank you for the keynote this morning. Great. Excellent, thank you. And this has been another piece of their journey, and we'll see each other in two weeks. listener if you haven't subscribed yet, you can find this podcast on iTunes, Stitcher, Google music and much more. And if you like what we do, please help your fellow developers discover the podcast by writing it and writing a comment on those platforms. Thanks again and in two weeks.