The past decade has been a period of incredible growth for the technology industry in Seattle and the Pacific Northwest. One of the primary engines of that growth is the University of Washington’s Paul G. Allen School of Computer Science & Engineering in Seattle. Computer scientist Hank Levy has been a quiet force behind the program as its leader for the past 13 years.
RELATED: Harvey Mudd President Maria Klawe has a message for computer science grads — and the tech industry
During his tenure, the program positioned itself as one of the top 5 computer science programs in the country, after MIT, Stanford, Berkeley, and Carnegie Mellon in the minds of many in the industry. It grew its faculty by 30 positions, or 70 percent; doubled its space with the addition of the Bill & Melinda Gates Center for Computer Science & Engineering; tripled its undergraduate enrollment and doubled its graduate enrollment; and developing strengths in areas such as robotics, data science, security, sensors and machine learning.
Levy will step down as Allen School director effective July 1 but will remain involved with the program. Speaking with GeekWire this week, in advance of the Allen School commencement Friday evening, Levy reflected on the huge changes during his tenure, the transformation of technology during his career, and the challenges still facing the UW and the tech industry.
Listen to the conversation as a podcast below, or subscribe to GeekWire in your favorite podcast app, and continue reading for an edited transcript.
Todd Bishop: Hank, it’s great to have you here.
Hank Levy: Thanks Todd. Thanks for inviting me. I really appreciate actually all that you guys do for Northwest tech and for promoting the great work we’re doing in the Allen School as well.
TB: I’ve got to say a lot of times we’ll sit around here at GeekWire and say, “We’ve got to get over to the UW. We’ve got to get over to the CSE program more,” because every time we go over there it feels like we’re seeing a glimpse of the future. Because your students are working on sensors and wireless technology and back end infrastructure, the things that not only provide the plumbing for everything that we use in our daily lives, but some really advanced technologies. I’d be curious just to start this conversation here to get a sense for the state of innovation as you see it now from your vantage point. What’s going on out there that people should be paying attention to?
Levy: Actually, I was going to say, just to follow up on what you were saying, I feel the same way even though I’m inside of the Allen School. I’ll talk to people in the school and say, “What are you doing?” And they’ll tell me, I’ll just say, “You’re doing what? How is that possible?” So there’s a lot of really exciting work. One of those things that got a fair amount of publicity was ambient backscatter a couple of years ago. Shyam Gollakota, Josh Smith. Which is building small devices that are not, have zero battery power that essentially just pull power out of the air from radio signals. And when I first heard about that and the way they’ve advanced that technology over the years, it’s pretty amazing.
TB: It really is. I remember I saw the first demonstration of this, I think it was back at the old Intel lab where they were.
Levy: Yeah, that’s right. The lablet.
TB: Yeah, exactly. In the University District. They’re no longer there, but they had an open house. And this is fascinating technology because it can fuel something like IoT, the Internet Of Things.
TB: And the fact that you don’t need to have a battery there or a power connection is incredible. And as you said, they’re basically pulling power out of the air. That alone is amazing. But it just illustrates the fact that … It amazes me how we can take things in the world around us sometimes and create things that, I mean, who would have imagined this even 50 years ago?
Levy: Yeah, it’s great. And for IoT it’s really important because we want to have these little sensors and devices everywhere and you don’t want to run around your house replacing the batteries every hour, which is what you would have to do eventually when you’ll have a a thousand devices in your house.
TB: You’re involved in operating systems, distributed systems, the internet, computer architecture. Give folks a sense, because I think you’re kind of the quiet force behind UW-CSE as opposed to your colleague, Ed Lazowska, who I think a lot of folks know, who’s very much out there in the public. And it works well. You’re sort of behind the scenes.
TB: Explain how you got into computer science originally.
Levy: Boy, I was just very, very lucky. I got into Carnegie Mellon as an undergrad, not because I was a good student, but because the soccer coach got me in. And had had a little bit of, so I’m 66 just to tell you, so we’re talking about the early 70s. I was lucky in that my senior year in high school, there was a course sponsored by NSF that taught us a little bit of computer programming. But to give you an idea of what that involved, we had a card punch in our high school. We would write little programs, we would punch the cards, and Friday afternoons they would bus us to a small local college that had a small IBM mainframe, where we would read in the cards and discover that we made a mistake punching them. Which meant you basically went back and did it over next week.
TB: Oh my God.
Levy: Just to give you an idea.
TB: Back on the bus.
Levy: Yeah, you got back on the bus. So that’s what it was like programming. As an undergrad I actually worked for the computer center at CMU as a systems programmer and operator on an IBM mainframe. And similar, I would read in card decks at night and then I would actually have to load the disc pack so that that job could read its data. So I’ve seen a lot and actually it’s just been an incredible ride for computer science over the last 50 years. I’m biased, but I like to say that there’s never been anything like computer science. I don’t think in history there’s been anything that’s been on an exponential growth curve for 50 years without stop. And that gives you some remarkable things.
‘Almost nothing is similar’
TB: How much of that can you attribute to human ingenuity and how much can you attribute to just happenstance, that these things reached a confluence at this point?
Levy: Of course, a lot of it was driven by the semiconductor industry who was providing us with these devices. There’s what’s called Moore’s Law, which is not at all a law, it was actually Intel’s goal to deliver a doubling of transistor density on chip every two years or every 18 months or something like that. And they did that for decades and decades. And what that meant was people could build complex systems that maybe didn’t run on today’s processors, but within two years, those applications would be able to run. And that was a remarkable thing.
Levy: Of course, many people don’t know that Moore’s Law died about a decade ago and so-
TB: Did it officially?
Levy: Well, there wasn’t a funeral but, but basically what happened is that we have so many components on the chip that we can’t power them all. And so we’ve had to do other things. But that’s led to new excitement. And I think the bottom line is, humans are incredibly clever and incredibly good at problem solving. And so when you put one problem in their way, they come up with other ways of addressing that problem.
TB: They pull energy out of the air.
TB: So when you think back to your experience there in Pittsburgh at CMU, and then you see the environment around the students at the Allen School for Computer Science and Engineering now, those who are going to be graduating this Friday, how do you compare and contrast the two? What’s similar in their current environment and what’s different?
Levy: Boy, almost nothing is similar. I like to talk about computer history when I’m teaching because I want them to not take for granted what we have. I also, actually, before I came to University of Washington, I spent 10 years working at Digital Equipment Corporation, which was called DEC, was the leading minicomputer manufacturer. And actually at one time it was the second largest computer company in the world. The fact that most people never heard of it because it doesn’t exist, is an indication of what happens when you lose sight of technology and how quickly it’s moving. And that’s what happened to DEC, in part.
Levy: But our students live in a remarkable world where technology just surrounds us, and everybody is affected by it and everybody knows they’re affected by it. It’s not just computer scientists. I think one of the things that’s amazing is, again, it goes to the rate at which computer science is improving. So if I look at the curriculum, the way I look at it is somehow every, let’s say, five or six years, we’ve taken what used to be graduate work at least at the Master’s level and shoved it down into the undergraduate level. And our undergraduates can absorb that because they have so much background. What’s amazing today is not only how much they learn and how much they can do, but how much students coming out of high school already know.
Levy: So when I talk to students who are applying to get into our program about what are they interested in, they’re not saying, “I want to do computer science,” they’re saying, “I want to do artificial intelligence applied to energy technology,” or something like that. They’re very, I don’t love this, actually, in that I don’t want people to be narrow. I want them to be broad. So sometimes they’ll say to me, “Oh, I want to do this little thing.” And I’ll say, “Kid, you’re only 18. There’s a big world, take a lot of courses, be broad,” because the excitement is at the intersection between different fields or different subfields. It’s too easy for people to be deep either before they come to undergraduate or before they come to graduate school. But at every level people are exposed to this technology and are impacted by it.
TB: It strikes me that one challenge with this current generation of budding computer scientists might be to make sure they don’t take the technology around them for granted. They understand that hey, this was created by someone else. It’s not just automatic and you have an opportunity to create the next thing that the next generation might take for granted.
Levy: Yeah, that’s true. I think people have to realize that, that they have an opportunity to advance the field and to put it in a better position and to solve really important problems. And then hopefully people can go from what they’ve done and continue to do that in the future.
TB: So when you look at the technologies that are out there today, and I don’t want you to prejudge the situation for a student coming in, but if you were starting over, just starting computer science school now, what would you focus on? Just based on your own personal interests?
Levy: You know, Ed Lazowska and I talk about often the fact that if we were applying now we would never get in. [laughter] So we were very lucky to come here early when nobody had discovered how much fun this is. I think one of the things that’s exciting to see, and we’ve really grown the Allen School in this direction, that’s one of the things that’s made us, I think, a leader nationally, but kids really care about impact and that doesn’t mean just technical impact. That means changing the world.
Levy: And we’ve used that, we care about that, as a program also. And we’ve grown the program in the way that we are very strong at core computer science. We are strong across the field, but we also have a number of people that are solving what I would call grand challenge societal problems. And the fact that you can do those things … Most of those problems, whether it’s in education or transportation or energy or healthcare and medicine, curing diseases, those are really important problems. We don’t have all the keys to those problems. On the other hand, those problems can’t be solved today without computer science.
TB: You didn’t answer my question.
Levy: Probably right. What would I do? That’s so hard. I’m old and I’m stuck in my ways. I still like operating systems and computer architecture, which is kind of the center that I came from.
TB: So would you probably, I imagine if you were to get back into operating, I imagine if you were to start over in operating systems, you’d probably be looking at some sort of cloud native architecture at this point.
Levy: Well, right. There’s, operating systems as I knew it was different, and the fact that cloud works is a result of a lot of research in part that went on, let’s say, in the 90s in my field, where people were looking at … the big issues in the 90s were scalability, how do you scale systems to larger numbers? And how do you make them reliable? And things like that are keys to cloud computing. But obviously, we’re very lucky to be here in Seattle, which is Cloud Central, where Amazon remarkably really made this work at scale. Microsoft is doing great work in cloud, and actually there’s a lot of work at Google in Cloud that’s going on right here in Fremont, a couple of blocks from where we’re sitting right now.
TB: Soon to transfer down to South Lake Union.
Levy: To South Lake Union.
TB: Yeah. But it is, it was actually fun yesterday, I got to interview Adam Selipsky, the Tableau CEO, about the Salesforce acquisition. And I got to walk down the Burke-Gilman Trail past Google down to Tableau. It can feel like a small town here at times.
Levy: And it’s amazing how new that is relatively, right? 20 years ago that was not Fremont.
TB: Exactly. And 20 years ago, if you weren’t working for Boeing, you were working for Microsoft. And if you were working for Microsoft, I would imagine the options for engineers here were far fewer.
Levy: Yes. Yeah, that’s true. I was first here in 1980 for a year actually. I spent a year essentially on sabbatical from Digital at the University of Washington and I got to see Seattle, which was a very sleepy town at the time. I describe it to people by saying that, when I first came here, you could go to any restaurant in the city on a Saturday night and get a table and sit down and have dinner without a reservation. Now you need a reservation but there are tons of great restaurants. And that’s part of the upside and the downside.
TB: This might be a slight digression, but we spent a month, last year, GeekWire did, in Pittsburgh for our own HQ2, which was part spoof, but actually reality, we were there covering Pittsburgh for a month. I got to say, I was so impressed by CMU and I know that there’s a relationship between the two, where sometimes their professors come to UW.
TB: But it was a great illustration of just how much a university can fuel a technology economy, a tech industry in a community. And it’s interesting, we came back here and one of the most instructive things about our Pittsburgh experience was just how amazing Seattle is by comparison. Pittsburgh is fantastic —
Levy: Right, but it’s not Seattle.
TB: It’s not. And the impact of the UW and the computer science program here, I think, is sometimes taken for granted, because Seattle is so much bigger. What kind of relationship does the computer science and engineering program have with the broader tech community in Seattle?
Levy: Well, first I want to say, and you might bring me back to the question, but if you look at all the high-tech centers in the country, it’s pretty clear that they are close to, and they surround, top universities, and in our case, top computer science departments. It’s not an accident that Silicon Valley is next to Stanford and Berkeley. It’s not an accident that the high-tech region in the Boston area is near MIT and Harvard, and many other schools. Even in North Carolina, there is Duke, UNC, which had a very early program in computer science, NC State. So people purposely locate around universities because they’re a source of ideas, they’re a source of workforce in terms of the students that we graduate. They’re a source of consulting, and collaboration with faculty. So all of these things are there. People in industry want to be close if they’re going to grow, to a university for the technology that’s coming out of it, and for the people that are coming out of it. People are the secret. If you’re doing a startup, those people that you hire, those first people that you hire are everything, and the quality of those people. It’s harder to hire those people in the middle of a corn field than it is in a place like Seattle.
I don’t know if you know, but somebody did a LinkedIn study a number of years ago, of people who identified themselves as software engineers, and looked at where they went to school and where they ended up working. We had the highest percentage of software engineers who were educated in Seattle at University of Washington, and stayed in Seattle, compared to anywhere else.
Levy: It’s such a great place to live and great place to work. We benefited enormously from the tech community around us in many ways. But, the most obvious was, we have been growing incredibly in the program, we’ve tripled the undergraduate program. We’ve almost doubled the faculty. We’ve doubled the PhD program. We were out of space and we needed to build another building. That’s a challenging proposition, and we were able to do that, in part, from support from the state, of course, and the university. But, the majority of the funds were raised privately. That was because people like Brad Smith at Microsoft, and Microsoft corporation, and Amazon, and Zillow, and Google, were able to support us. Other people in the community, Charles and Lisa Simonyi stepped forward and really helped us. Just, many, many people. In the end there were something like 200 to 300 donors. But, to have those local companies come forward and say, “We really value what you do, and we want to support your growth,” was just an amazing thing.
TB: To me, the beauty of that is the long-term, decades-long, virtuous cycle that you can point to, where Bill Gates and Paul Allen, now, granted the university didn’t explicitly support this, but they would sneak in to the university buildings late at night, and learn how to program on punch cards, right?
Levy: Yes. I think you were there when Paul Allen, when we had the 50th anniversary and we announced the creation of the Allen School. Paul Allen actually brought with him, the letter that had been written to him however many … in the ’80s, from Hellmut Golde, who was then the head of computer science, telling them that they had been illegally using resources in the computer science department, and that they committed this sin, which is they borrowed an acoustic coupler, which was a device we used 50 years ago to communicate from a home computer to a big computer somewhere over phone line.
TB: It was the Internet.
Levy: Right, it was a point-to-point Internet, that they had borrowed this acoustic coupler “without permission” from the professor, and that they were to return the coupler and to cease using department resources.” It was just so funny. One of the things I did after that was, I went to eBay and I found a 40 year old acoustic coupler. We packaged it up with a letter from Ana Mari Cauce our president, saying, “Dear Paul, we’re very sorry that we did this in the past. To make it up to you, should you require it, here is your very own acoustic coupler.” Of course, Paul loved it, because he loves computer history.
TB: So I imagine that might have ended up at the Living Computers Museum?
Levy: I’m not sure it was in that condition.
A ‘feeding frenzy’ for AI talent
TB: Oh, man, that’s great. Yeah, that was actually, in hindsight even, more notable because, of course, Paul Allen passed away last year. He and Bill Gates have both been major supporters of the UW computer science program. But, obviously, it is symbiotic relationship between the tech community and the university. But, over the past couple years, I know that some companies have brought your professors onboard. Correct me if I’m wrong here, my outside perception as somebody who pays attention to these things is, it’s caused a little bit of conflict that wouldn’t have been there in the past, with companies like Facebook, for example, poaching, maybe that’s too strong of a verb.
Levy: That’s a perfect word.
TB: Correct me, fill me in here.
Levy: Well, first, I want to say, well, you asked me before about the relationship, and one of the interesting things about computer science, I mean, I came out of industry, became a professor, which was an incredible thing for me. People go in both directions, and have always … There’s a lot of fluidity, and has always been a lot of fluidity in people and ideas going between universities and companies. That’s a strength of the field, because when I came here from DEC in those days, I brought a lot of experience that was really irrelevant to both my teaching and my research. So spending time in industry, and seeing what problems they face, is really valuable for people who are faculty members.
OK, that said, there is kind of a feeding frenzy right now by industry, particularly in AI, because everybody’s realized how important AI is to their future. There is not an abundance of AI people just sitting around at the bowling alley, if that even exists, where you can go and say, “Hey, do you want to come work for us?” So people are feeling this extreme need because it has had an impact on every part of every industry, not just ours.
So they are doing that, but you talked about the long term vision, we need those people to produce the next generation of computer science students, who are AI experts, and who they can hire. You kind of can’t have it both ways. This is not just our problem, this is a national problem for computer science. CMU, actually, has faced this. Uber hired 25 researchers, not necessarily to tenure-track faculty members. But, 20 or more people out of their robotics lab, that got national press.
TB: Self-driving cars, the lab there.
Levy: So this is going on nationally, and we have to figure out how to live in a world where people have maybe joint relationships, but it’s difficult. We can’t survive if every faculty member is half the time somewhere else.
TB: But, at the same time, you’re not blessed with unlimited funding from the state. I know this has been a long time discussion, but you’re competing with companies that have huge amounts of money to bring in top computer scientists. How do you do that?
Levy: There’s always been a differential, but we’re in an extreme, where you’re really talking about football player type salaries for very top people in some of these high demand fields. It’s hard to compete with that. I think industry’s going to have to help us figure out how we survive, and how we educate the next generation, and how we do the research that will help them, as well. I don’t think we’ve figured this out, but we need their help and their understanding to do this.
TB: To use an agriculture analogy, they’re taking the seed corn.
Levy: Yes. That said, again, we value the relationship with industry. We value their support. We want this community to be stronger, and sometimes that takes research. An example is that Dieter Fox, in our department, who’s a roboticist, was contacted by Nvidia, who wanted him to come to California. Instead, what’s happened is that they’ve opened a new research lab in Seattle, under his guidance, so he’s taking a leave. He’s set up new Nvidia lab. That lab is focused on robotics. He’s in his second year now, he’s going to transition back to UW. So that’s a win-win. There’s a new resource here in walking distance from us, with whom we’ll be able to collaborate in the future on different kinds of hardware technology and robotics technology, and we have a strong connection.
You mentioned the Intel lablet that used to be here. That actually had an interesting model, where that lablet was run always by a professor in CSE, I don’t know if you know this, on a three year rotating basis, and we supported that. Someone would go there, they would kind of move the lab a little in the direction of the research they were interested in, and they would do that work with students. There was a very nice open IP agreement, so students could go over there, and spend time there, work on things there. Jointly, they had access to Intel technology and Intel staff. Then, they would come back, bring back what they know, be a professor in their department. Someone were to go over, another person would go over for a couple years and move that lab in a different direction. We really liked that. I thought that was a great opportunity and a great relationship with Intel. We were sorry that they closed down eventually.
TB: Why haven’t more companies done that? Because it seems like that would be more of a win-win?
Levy: We always give that as an example of when companies want to know how to collaborate with us. But, I think that’s hard for lots of companies to actually … It’s an expensive model, obviously. You open a lab, you hire a bunch of people. So I thought it was great that Intel did that. There were only a couple of those in the country, at places who they wanted to collaborate with. There was one at Berkeley, one at CMU, one at UW. Those were the lablets, and eventually, somebody decided that that money could be better spent.
Diversity in computer science
TB: My colleague, Monica Nickelsburg, had a great conversation with Maria Klawe, who is giving the commencement address at the University of Washington Computer Science and Engineering Allen School graduation. She is the President at Harvey Mudd College. They are known in part for the fact that 50 percent of their graduates are women. I know the University of Washington is ahead of other universities, but it’s not quite at that mark. Where are you now, give me your assessment after 13 years of this entire issue, if you don’t mind here?
Levy: In our undergraduate program, we graduate about 30% women, which is much higher than the national average. I think the national average has improved, it used to be 15%, maybe it’s closer to 20% now. But, we’ve done very well here. Our graduate program’s probably on the order of 26%, something like that. So we’re kind of known for this, and we’ve worked hard, many people have worked hard on this.
It’s hard to attribute our successes to a specific thing, because we do so many different things, and it’s hard to measure. But, we do lots of outreach to Seattle school system K-12, high schools, we have summer camps for middle schools, particularly for young girls and boys, but focused often on girls also, to show them how exciting this is. We bring people here and give them demos of exciting work. We do all kinds of things to reach out. It turns out that our intro programming course, which is enormous now. We teach on the order of 4,000 students a year in various intro course, particularly the first two intro to programming and some others. But, the first course is like 40% women right now.
So the great news is that everybody realizes that computing is important to their lives. More and more they realize that it’s not just about video games, but it’s about social networking and communication between people, and it’s about changing the world, as I said, and solving really important problems. Students are being introduced at a young age to programming, so it’s not unusual for them. In the home, everyone has a smartphone, so they’re aware of what computing is, it’s part of their lives. They realize, “Oh, I could do some of this.” So I think that the women are there. We have to convince them, because there are lots of good things they could major in, that they should major in computer science, and we work hard to do that as well.
TB: Have you started to see students at the undergrad level who were introduced to programming through initiatives like Code.org or other things like that at the high school level, where you’ve seen this new push over the last decade to make sure many students are introduced to the field?
Levy: I don’t have data on things like Code.org, but more and more they are introduced to it in high school and they take AP, and then they actually place out of the first course. But as I said before, we are taking, this year we’re taking 50 percent directly out of high school, and then we bring those students and their families here to try to recruit them because they have lots of good options, and one reason for doing that is to keep the good students who are local, local, rather than having them go to the obvious other places.
And lots of the young students, but including the young women who were there, as I said when I talked to them … Actually, one which made me feel really good just sent me an email telling me that she was coming, and I just happened to see her and her dad in the hallway during this visit. And I said, “What are you interested in?” And she said, “I’m interested in robotics.” And I said, “Oh, come with me,” and I took her to the robotics lab and showed her all the incredible stuff there, and I could see her eyes kind of open up. Then she sent me an email just a couple of weeks ago saying just want you to know that I’m coming to UW, so that was a great thing.
Levy: But again, at that age, it wasn’t just computer science; she was interested in robotics. Another was interested in environment, as I said, with computer science. They have just absorbed the fact, kids at all ages, all genders, that technology is part of their lives and that they can use it to do what they want.
TB: I know that a lot of parents out there would want your tips, and kids would want your tips, and I actually can personify this. I have an 8-year old daughter, wicked smart at math, zero interest in math. What would be your advice to somebody like me with a kid at either grade school or high school, how do you introduce a kid to computer science in a way that doesn’t alienate them from the field?
Levy: I’m embarrassed to answer this since my kids have no interest in math despite the fact that I’m a computer scientist and their mom is a biostatistician, so you have to kind of go with what they’re interested in. But you have to find an ‘in,’ and maybe that’s a summer camp in something that interests them, that shows them an application. But computer science, and math and statistics, that intersection is where it’s hot right now, particularly statics and computer science because that’s machine learning and that’s artificial intelligence.
Levy: So increasingly, everybody will need those skills if they’re going to have not just high tech jobs, but outstanding jobs, because data is everywhere. Everyone’s collecting data; everybody’s analyzing data; everybody’s trying to learn from data. And so they have to have an understanding of these fundamental technologies that underlie that, whether it’s how do you visualize data, how do you store and query data, how do you learn from data. People need those skills. That’s why we call it data science globally, and there’s an effort at the University of Washington that every student should understand data science and have some skills in data science because it’s important for them no matter what field they’re in.
If you’re in sociology or if you’re in psychology, you are going to be exposed to data, and large amounts of data, because you can collect it online; you can collect it from censors, and you need to understand how that data is analyzed. If you’re in business, you need this; if you’re in medicine, you need this, so it’s important for everyone.
TB: One of our most loyal event attendees and readers is Ben Slivka. I would be remiss if I did not ask you for your formal title so we can get that on the record.
Levy: Yes, I am the Wissner-Slivka Chair in Computer Science & Engineering at the University of Washington, and Ben and his wife, Lisa, have been great supporters to us and we really appreciate it. And he’s just an incredibly fun guy who comes over and talks to students, and gives them feedback, and always appears in shorts and a Hawaiian shirt, and that makes him a lot of fun.
TB: Ben was at Microsoft for many years.
Levy: Yeah, and then at Amazon, also.
Diving back into research
TB: Oh, yeah. That’s right. So catch people up on exactly what you’re doing. You’re not technically retiring.
Levy: I’m not technically retiring; I’m just stepping down. I like to say I’m in the 13th year of my five year term as head of the program, which is what it’s supposed to be. And why 13, I think I can say that I was not good at arithmetic and could not count past five, but various things happened along the way. One was, the recession happened. I start in 2006. There was a recession. That was a tough time. We had a change of deans, and having strong leadership on the team for the Dean of Engineering was important.
Then we started working on the new building, the Bill and Melinda Gates Center, and it seemed like changing horses was a bad thing to do when you’re trying to raise $100 million, so all these things happened. But anyway, my goal is to go back. I am really … The organization has really grown so big. When I first started, I didn’t think of it as being a full-time job and I was still very engaged in research and teaching, and I still have some PhD students, but it’s been hard over the last eight or nine years to keep up. So I want to go back and bring myself back up to speed technically.
TB: So my question for you at the beginning about what you would do if you were starting over again was not actually a hypothetical?
Levy: No, no, no.
TB: What are you going to do?
Levy: Well, we’ll see. I might spend some time at one of the companies locally. I might sit in my office, close the door and read lots of papers. I’m just at the point of figuring that out right now.
TB: Oh, that’s very cool. Very cool. So big picture thoughts, when you reflect on your tenure as the head of the department, as the Chair and now Director, what would you want people to know? What’s the most important message you would give folks?
Levy: It’s been an amazing time, and I feel so lucky to have been allowed to do this job and to work with such great people. It’s an incredible organization and we have a reputation for being one of the most collegial computer science schools, departments in the country, and so it’s really fun working with these people.
We’ve done something that I think hasn’t been done before. Kind of forever, there were four number one departments: MIT, Stanford, Berkeley, and Carnegie Mellon. My goal coming in was actually to change the landscape so that people thought of it as a top five, and I think that’s happened. We’ve clearly pushed into that group and it’s very exciting. It’s also very challenging because the competition is those people, and the kinds of resources they have are phenomenal. I don’t know if you know, but MIT, in part seeing us become the Allen School above many other things, decided to create a College of Computing and AI. And the investment is, are you ready? $1 billion.
TB: $1 billion. How’s that compare to annual funding for …
Levy: For the entire University of Washington? And that includes 50 new faculty positions approximately.
Levy: So that’s … And Stanford has also putting $1 billion into an AI organization. So those are just the kinds of numbers that are unheard of in the past, so that kind of competition is almost impossible. But anyway, the important thing is that we have moved to become really a national resource, but I want to say again how much we appreciate being in Seattle; appreciate the support of the University, of local tech industry, and actually of the Washington state legislature. I hope they take the same pride that we do in the organization because to a large extent, they’re responsible. They recognized the importance of a strong computer science program at the core of a tech industry, and they were willing to fund our growth. And in response we, as I’ve said, tripled the undergraduate program.
TB: I was referring earlier to that long term virtuous cycle, and when you think about that and then you put into the equation what’s happening at MIT all the more reason for even these companies that are coming up from Silicon Valley and establishing engineering centers here to step up and contribute, and not poach in a way that diminishes what you’re doing.
Levy: Yeah. Well, we have a great … despite the … poaching is a pejorative word, but we have a great relationship with all of these companies. And one thing to know is that increasingly, we need to have relationship because some of the resources we need cannot be found locally. For lots of AI research, for example, you need access to special hardware; these are called GPUs or TPUs, or whatever. But they’re special purpose hardware devices that are used to accelerate machine learning, training in particular, and other things.
If you’re at Google and you want to solve a hard problem, you can say I need 50,000 GPUs. We don’t happen to have, we don’t have enough power on campus to have the numbers that these people have. And so having access to that, having access to data, and having access to the problem set, understanding the problem set of industry is important to us, so we want to have a relationship with these companies, and we do. But there has to be a balance, and we need to be able to continue to do research, hire good people, keep those people, and to teach our students the skills that industry wants.
TB: Well, Hank Levy, best of luck, and thank you for being here. And I’m looking forward to seeing what you do after you close the office door and figure out what’s going on in those research papers.
Levy: Thank you, Todd. That was fun. I really appreciate it.