Treehouse for Teams
writes on October 15, 2018
Robert Krohn is SVP of Engineering at Pindrop. He has over 20 years of engineering, management and executive experience in software, cloud and operations across diverse product segments at both startups and public companies. Robert was most recently the VP of Virtual Managed Services (VMS) Software Engineering Group within the Network Function Virtualization BU at Cisco. Robert is a graduate of RMIT University in Melbourne, Australia.
Subscribe to Change Wave on iTunes, Google Play, Spotify, Soundcloud, Stitcher or your podcatcher of choice.
Contents
Show notes:
Robert on LinkedIn
Transcript:
Ryan Carson: Welcome to Change Wave, an exclusive look at the real, first-hand stories of how cutting edge leaders rose to the top, smashed through barriers and created real change. I’m your host Ryan Carson, the founder of Treehouse, the company that’s taught 850,000 people to code. We also help companies like Adobe, Nike, MailChimp, Airbnb and more hit their hiring plans and create diverse teams. If you’d like to know more, head to teams.teamtreehouse.com/talentpath.
Today I’m joined by the illustrious Robert Krohn from Pindrop. Thanks so much for hanging out.
Robert Krohn: Oh, hi. Great to be on the show.
Ryan Carson: I decided to give you the illustrious title because I thought that would be fun. Why don’t you give us your official job title though?
Robert Krohn: It’s actually SVP of Engineering at Pindrop and I report to the CEO. I run all of engineering for Pindrop.
Ryan Carson: Awesome. Tell us about Pindrop. What do y’all do?
Robert Krohn: Let me give you a bit of background because we believe that actually voice is the interface of the future. People will be really interacting with their devices, really with voice. Today we help large enterprises in their call centers deal with two problems. One is how to authenticate their users as they call in in a really easy way that seamless and fast. And also how to catch fraudulent callers. And we do that with really advanced audio analysis and machine learning techniques. So it’s a very interesting approach to authentication and catching fraud.
Ryan Carson: Interesting. My wife noticed something … And I don’t know if this is related at all to Pindrop but I’m curious if it is. We started seeing on our iPhone when a scam call would come in it started to say something like potential fraud on the id. And I thought oh, that’s interesting. What’s kind of going on with that behind the scenes? Do you know?
Robert Krohn: Yeah. We have some techniques in that area as well actually. When robo callers call in or scammers call in to your phone, they often do something which we call spoofing the caller id. They actually change the caller id to match something that is familiar to you. Like in your home area. So you often see calls that are coming from your area code and even the first three numbers actually, they can spoof. So you think oh, it must be the neighbor and you pick it up. And you can detect that and that’s sort of what’s going on. Some of the carriers have started to work on that. And we’re actually working with some of the largest carriers in the US to add that feature to determine as a call comes in to a call center, what is the risk associated with that call and if it’s being spoofed it’s a very high correlation that it’s probably a risky call.
Pindrop’s org chart
Ryan Carson: That’s great. Nice. I think we’re all for this technology. It’s nice. You already explained. I always ask people, how do you fit in the org chart? So it’s kind of interesting you report directly to the CEO. What is the rest of SVP kind of structure like? If you want to share.
Robert Krohn: Sure. We have leaders that look after marketing, products, also customer success, and we have a CFO and a CEO. Actually our CEO, Vijay, is both the CTO and the CEO. So at the top it’s a fairly flat, smaller structure. We’re about 300 plus people. Mainly in the US and Europe. So we want to keep the management structure fairly coherent and not too broad.
Ryan Carson: Got it. It’s interesting to hear as well how sometimes engineering and products sometimes report to the same person, sometimes they’re separate. So it sounds like product management, product design reports in to someone else.
Robert Krohn: Yeah. It reports in … There’s one VP that’s responsible for product and we sort of incorporate a little bit of UX as part of that. But you’re right. That’s actually an interesting phenomena. Do you have product and engineering at the same level? Do they report in to the same person? That’s a very interesting dynamic that goes on. And ultimately you really want them to be balanced. There is advocacy in both sides. Product is an advocate for the customer and the business. And engineering is an advocate also for the customer but it also has sort of internal things that it has to take care of. Things like architectural improvements and technical debt and things like that.
Ryan Carson: Right. Yeah it’s interesting, at Treehouse we have our Chief Product Officer, she’s also essentially our SVP of engineering. And it’s interesting, it’s like that just kind of made sense for us and her certain skillset and experience but lot of other people I talk to don’t do that. It’s interesting to see how it differs. So let’s dig into a little bit more about you. You talked about the engineering team but how many directs do you have?
Robert Krohn: I have about seven directs. Pindrop is an interesting company in that we cater to some of the biggest enterprises in the world. Some of our customers are banks that have to protect assets like accounts and stuff like that. So we have not only engineering that goes on in engineering but we have some engineering and research that goes on outside of engineering. So we have a large research team and we have a large customer success team. So although I have all of engineering sort of in the traditional sense and I have a bunch of directs that are looking after functional areas within it, I also have to interface with my research pals who are doing research in algorithms and some amount of engineering to package that stuff up. As well as the customer success guys who are doing sometimes things like professional services or configuration into customer call centers. So six, seven directs for me but I wanted to point out that the development at Pindrop isn’t just limited to my organization.
Mentorship
Ryan Carson: Got it. Thanks for sharing that. I’d love to know more about your favorite book or course or person that’s influenced you as an engineering manager and then ideally what was kind of one of the big takeaways that you took from that source?
Robert Krohn: I’ve been really lucky in that I have had so many mentors throughout my career. I arrived in Silicon Valley in 1990 and I’ve worked at big companies and small companies and startups and I’ve started my own company before. And along the way I’ve just learned from some amazing people. And I’ll give you some examples of that. But even at university I started to learn about software engineering. I went to this place in Melbourne, Australia called RMIT and I did a software engineering course there. And I even learned stuff there about how to make projects work well, how to interact with other teams. And then the other big influencer for me, which I’ve come to realize in the last 10 years or so, is my father.
Ryan Carson: Interesting.
Robert Krohn: He taught me a lot of stuff around work ethic but one thing that he said to me over and over, he said “There’s no such thing as I can’t.” Whenever I would say oh, I can’t do that or I can’t do that, he would just stop me and say “No. That word doesn’t exist.” And I’ve recognized that that sort of come into my work practice where I question my team to say “Why can’t you make that algorithm go faster?” Or “Why can’t we make our CICD systems go quicker?” You know “Why can’t we find that bug?” I ask a lot of questions because I don’t want to give up. And i realize that I got that from my dad.
Ryan Carson: That’s interesting. What kind of reaction do you start to see from your team when you start asking that? Then what happens that’s beneficial?
Robert Krohn: You’ve got to do it in the right way. You have to do it with a smile. You have to have curiosity. You need to be careful of not providing a possible hypothesis or answer too early. Let the team sort of work out what could be the root cause. But just probe more and help them get to that conclusion. I try to be positive. I try to be encouraging and be helpful when things are difficult. Sometimes the results and the solutions come from unexpected places. I remember once we were working on a product and it had this particular set of metrics that weren’t that great. So it was like calls per second. Let’s just make that up. It was doing like 1,000 calls per second and we really needed to get to 15,000 to 20,000 per second. It was like a web API. And we were sort of trying to work out what the problem is and this one guy put up his hand and he was a young engineer, and he said I think I found it, after a few days. And he just quietly worked out what the problem was and didn’t give up and actually got this web API to go from about 1,200 calls per second to around 25,000 per second. I use that … Don’t give up. Keep looking and often the answer is in front of you.
Ryan Carson: It’s cool that your dad taught you that as well. Since we’re both dads it’s kind of encouraging to know we can have a lasting effect on our kids.
Robert Krohn: Yeah, exactly.
Vital lessons for climbing the management ladder
Ryan Carson: If you could go back in time to the younger Robert and sort of grab him around the shoulders and make him listen to you, what would be two of the lessons that you would like to explain to him?
Robert Krohn: I think earlier in my career I didn’t spend enough time honing my practice around communication. I was a little quieter. I was a little more withdrawn. I hesitated getting up on stage and as I grew in my career I realized that that was a really important part of being a leader. Getting up and interacting with people, communicating over and over again and that’s one of the pieces of advice I give to engineering managers all the time. Over communicate. Over communicate the status of your project, where you’re going with it, if you’re having any trouble with it, any of the issues or bugs or concerns, over communicate about that. Set expectations clearly. So communication was something that I learned later in my career and I wish that I’d done more of it when I was younger.
Ryan Carson: Just a pause on that. For those people listening that are younger and are trying to go up the ladder, is there a particular book or podcast or place they can go to learn some of those things? Or are they just going to have to do it the hard way?
Robert Krohn: I think there’s plenty of examples. I mean how to use PowerPoint effectively. There’s philosophies about not using it or using more pictures. When you’re interacting with people, how to connect with them, how to read their body language, all that stuff is important. When you’re up on stage, not pacing around too much, using your hands appropriately. So there’s lots of things that you can learn there. But like any skill it’s about practice. And I think what’s really important is just doing it. What I do, sometimes I give large presentations, and I still get a little nervous about it, I do get the butterflies and I work on techniques that help calm that down and one of the techniques that works for me is to get up early and just crack a joke. Crack a joke and sort of break the ice and just get going.There’s plenty of materials out there but practice it. Get up in front of your colleagues. If you’re at school, give presentations. Go look at things like toastmasters. That’s another technique. Perhaps engage in English style debating. There’s lots to do but the most important thing is practice.
Ryan Carson: Got it. Great. Lesson one is improve communication sooner. That’s what you’d tell your younger self. Can you think of a second important lesson?
Robert Krohn: Yeah. Perfect is the enemy of good enough. When I was younger I strived for perfection. As you go up the curve of getting to the perfect algorithm or architecture or system or UI or experience, you get to diminishing returns pretty quickly. Sometimes you just have to get to “good enough” so that you can do more than just sort of auger down on this one problem. And there’s sort of a similar thing that I talk to my teams about is let’s say you have an existing tool in place. Let’s say it’s like a message bus or it’s a JavaScript framework or something. And often teams will come to me and say “Hey, we need to replace this. We don’t like it. It sucks. It has all these problems.” And my question to them is “Okay, I’m supportive but tell me about how this new framework or this new message bus or whatever is going to be better.” And what I look for is it’s got to be five times better in some metric. It’s got to be five times cheaper or faster or easier. And I’ll settle for two to three. I’ll start at five but I’ll settle for two to three. So I think about that a lot because there is a cost in replacing an existing system with a new one. No perfect architecture. There’s always things that can be improved and it speaks to continuous improvement as well.
Research
Ryan Carson: Gosh, yeah. That’s a great lesson. Thanks for sharing it. I understand you have something called Pindrop Labs, which might be separate from the general engineering teams. Can you tell us, if that’s the case, more about it and why you do that?
Robert Krohn: That’s part of our research team. Our research team, we do two forms of research. We do sort of more applied research where we’re looking at our algorithms and sort of let me talk a little bit about some of the actual technologies we use. One of the things that … We have two main pillars of technology that help us do that analysis that I talked about before and that is listening for the real caller or identifying a fraudulent caller. So two pillars there. One is something we call phone printing. We actually take a fingerprint of the phone that you’re using to make the call. And it’s unique. So we can do a match of as that call comes in we listen to the audio and say “Ah, we’ve heard that phone before.” And we extract about 1,500 machine learning features from that audio stream and then we do a match against either someone that has called before and we’re doing authentication, id, and identity match. Or maybe we’re matching them against hundreds of fraudsters that have called in before. So there’s phone printing and then also voice biometrics. And we have our own deep neural net voice biometrics system called Deep Voice. So there’s an enormous amount of science and machine learning and audio analysis that goes on there. So our Pindrop Labs, our research teams, are working on both how to apply in our customers environments, how to tune and how to get the best results out of these algorithms in real time and how do we develop the next generation types of techniques and algorithms for things that are coming down the road? For example, I talked about voice being the next interface. One of the devices that we believe is going to be very important in the future is things like home automation and smart speakers and even things like security systems and cars and things like that.
Today, the smart speaker market is growing faster than anything that we’ve seen before. But it lacks security and it lacks identity or authentication. So you can turn up the music, you can go and set your thermostat. You might be able to buy something from Amazon in an easy way, add it to your cart. But it’s not ready for any real secure transactions because it doesn’t have security and it doesn’t have authentication. Anyone can talk to your Alexa. So we’re working on algorithms that solve those sort of problems. For example you asked me to get better audio quality to put earbuds on. Well, that’s not possible with a device like that. You’ve got to deal with background noise. You’ve got to deal with voices that age. You’ve got to deal with when you have a cold. And at the same time you’ve got to be able to have a secure transaction that’s accurate and really works in very small segments of speech because the transactions that we have with these speakers are very short.
Ryan Carson: That’d be amazing.
Robert Krohn: That’s the sort of stuff that our labs are working on. Stuff that is now sort of in demo and prototype form and then next year or the year after we’re bringing out to market.
Ryan Carson: Boy, that would be so valuable. Will you be able to use it for things like logging in?
Robert Krohn: Oh, totally. Yes. I mean there’s a number of different use cases. So there is personalization. Personalization because it knows who you are. Because it’s about authentication and identification. It’s about secure transactions. And it’s also about security. Making sure that they’re not the wrong person. One of the things that we think is very important is we have to be able to deal with things like synthetic voice or like a replay attack where I record my voice and then use it later. So we detect all those things and that’s part of the security with automation that we’re working on.
Ryan Carson: Wow. The world would be definitely nicer when this rolls out everywhere because this is a very first world problem to have but I was at a beach house last week and I was trying to use Amazon Prime Video on the TV and I had to log in and then authenticate on an email and then enter in a pin and I’m like this is just crazy.
Robert Krohn: Usernames and passwords have to go. We need to replace them with other forms of authentication that are really more universal. We see a future where it’s your voice that is going to be used everywhere you are. You go home, you get into your car, you go to work. It’s a very, very accurate mechanism of determining who you are.
Ryan Carson: That’s exciting. I’d try to say a phrase to unlock something instead of having to do the madness that I was doing to just watch a video. That’s exciting. Going back to Pindrop Labs, you’re a large enough company that you’re able to fund that. I would like to know a little bit more about the moment where you decided to create a research team because you were able to do it. A lot of companies want to be able to do that and they’re not sure when, okay when do we allocate resources to R&D versus just pure building features? Can you walk us through a little bit on that?
Robert Krohn: In the case of Pindrop, it started from the very beginning. Our founder and CEO, Vijay, he actually did his PhD thesis that then turned into the ideas that formed Pindrop. So from the very beginning he recognized that we needed to have a blend of engineering and of research to make this cutting edge stuff work. It’s way too complicated to not put effort into it from the very beginning.
Ryan Carson: Right. You’re not building a simple web app here.
Robert Krohn: This is very sophisticated stuff. I think we have over 30 or 40 PhDs in the research group. So it’s a high caliber of people that are working there. My advice to companies, you need to sort of think about the domain that you’re working in. So if you’re working in say a web app or a mobile app you may not need as much of the sort of research that you expect. But one of the things that I did early in my career is I worked at a spinoff from Xerox Parc. And I was very lucky to be working at the Parc campus alongside scientists. So I’ve sort of been working with scientists and researchers from the very beginning in the early 90s and I recognized the value that they bring and they think differently to engineers. And in fact at Parc a lot of the user experiences that we were working on there, different ways of interacting with computers, didn’t come from sort of classical research scientists, they actually came from psychologists. Behavioral psychologists. So we had the very interesting approaches to how to interact with computers from different elements of research.
Diversity
Ryan Carson: Right. Because it’s a fresh perspective. Wow. Thanks for sharing. I want to dig in a little bit to diversity, inclusion and equity. It’s a subject that we’re all prioritizing and passionate about. I was just curious is there anything that you’re doing at Pindrop that is working that you could share with the audience?
Robert Krohn: Yeah, so diversity across all forms is super important and we’ve seen in the last few years, some real problems, especially in Silicon Valley around that area. One of the things that Pindrop worked on very early in its formation is let’s make sure that we have the right culture from the very beginning. And part of that is looking at diversity. So as we go through the interview process, as we go bring people on board and stuff like that, we keep that in mind and we actively seek to make that a balanced place for us to work.
Ryan Carson: Got it. All right. Thank you. We’ve gone long. A couple questions. I want to let you get back to your important work but I have a couple more questions for you. In case people that are listening that are engineers or are actually looking for a great company to work out, how many people are you hiring in the next 12 months and where do people go to learn about that?
Robert Krohn: We’ll be probably hiring around 30 or so engineers in the next month and we also hire … Let me give you some of the stats. We’re a company of about 300 plus. We have, I’d say, around 150 developers when I include the research team and some of the professional services team. And my team, we have about 100 engineers. In the next 12 months we’ll probably add another third. As we bring out new products and go into new areas it’s always good to bring in more people and also as we start to make our footprint in terms of products in other countries. So we have products today in the US and the UK. We also are moving into other parts of Europe and Asia. And it’s good to have people in local timezones for that.
Ryan Carson: Got it. Okay. Great. Let’s finish up with where people can find you online. Where’s the best spot to connect with you at?
Robert Krohn: Pindrop.com. That’s to look at where we’re hiring and then if people want to contact me personally they can go to LinkedIn and connect with me. I’ve got a pretty prominent profile there.
Ryan Carson: Perfect.
Robert Krohn: Easy to do.
Ryan Carson: Sounds great. And for you listeners out there, you spell Robert’s last name K-R-O-H-N. So just search for him on LinkedIn. Robert thank you for your time. Thank you for sharing some of your experiences. I especially enjoyed learning about the two things you would tell yourself if you went back in time. It was poignant. So thanks for sharing that and can’t wait for the future to be a place where we unlock things and log in with our voice so thank you for all your hard work on that and hopefully we’ll speak to you soon.
Robert Krohn: Thank you very much.
Ryan Carson: All right. Take care.
Robert Krohn: Bye.
