Luke Dai
Alright let's start off with some background info about you: what's your name, what's your year, what's your major, where you from?
I'm Luke. I'm currently a senior, and my major is Computer Science and Statistics minor. I'm from New Jersey. I've been here since like first grade; I lived in China a quarter of my life but yeah I was born here, moved to China, came back when I was in first grade.
Why or how did you choose your major?
It kind of started with my involvement in an extracurricular activity in robotics. I wasn't part of my school robotics because in our district there wasn't as much experience in terms of computer science or that kind of stuff, it wasn't as good quality as other districts in New Jersey and especially compared against California, so I started with doing robotic soccer, so we would have two or three people on a team and I would be the lead programmer. I would literally do all the programming and then my partner would be building things out of Legos and then we kind of created two robots that would do one kicker one goalie kind of situation, but after a while we went to internationals and we realized that the whole field in terms of US participation was pretty subpar compared to Germany or China, because they were using metal and had a lot more DIY stuff going on, and in New Jersey on our team, we only use, most of the US only uses lego. So that kind of turned me away from the hardware side of things and more towards the software side of things and that's kind of what really got me motivated to computer science as a whole. The second step would be the whole learning on my own kind of thing because it really doesn't have that environment. One opportunity I was able to get into was doing an A-star academy kind of camp in California, but other than that it was pretty limited, and so coming at Berkeley and doing computer science was kind of like the goal and what I wanted to do, at least being in California and being part of the computer science field. Statistics is kind of similar because I got into the neural networks, and you know deep learning, and so the next step up would be to you know actually learn about machine learning and stuff, so I took most of the 150 series and I'm currently taking more computer science classes, but I'm really in love with the intersection of these two things.
Is there a reason that you chose Statistics and CS as opposed to Data Science?
You know I wasn't sure about data science and when I was a sophomore data science just kind of popped out of nowhere so honestly, looking back, I felt like I would have liked it more if I did a Data Science major or something like that. It's still relatively new, there's a lot of amazing classes that are like I would say, now is the time to actually do data science in Berkeley because all these classes are being created and there are some really motivated people who are doing these kinds of class creations. It's different from these really long time classes like 151 or something like that, because sometimes you could teachers that don't actively try to improve the course; it's already fixed so there's no incentive to actually improve it, but in the data science field I'm feeling like these classes are being actively tweaked and getting better and better over the years, so that adds to the incentive, actually learning about you know data science as a whole.
In terms of your professional experience, what sort of stuff have you done the last few years?
Well I've been doing research for the past two semesters, I think, but yeah just in terms of neural networks, applying that to different fields. So recently I'm reading this paper with another professor on applying neural networks to survival analysis and so it's very important in the biomedical field, and a few weeks ago, actually Lucas and I, we were working with a grad student in her slot and we were doing coherence in terms of neuro natural language processing — I don't want to get into too much detail because it's not as important but yeah that's kind of things I've been doing in the past few months. And then in terms of internships I've done some internships in China with product companies that you don't know, as well as one with Uizard which has been our client for two semesters now and I'm pretty sure they're always welcoming towards doing another consulting project with SAAS. But it's interesting because Uizard was a startup that is situated in Denmark, so when I interned I was able to actually go to Copenhagen and live there for a month, and it was amazing. That all stemmed out of doing the first project with Uizard when I was in freshman year and Data Consulting and so that's why I kind of stuck with Data Consulting because yeah it's been a pretty amazing experience in this company. They're trying to create like a wizard for UI, so essentially instead of doing all the HTML, CSS, and JavaScript all you really have to do is create some bare bones outline essentially of what kind of UI you want, you take a picture and then automatically generate the code. It's all computer vision, the guy came out of grad school and then that was his paper so there's a lot of validity and this guy's like super smart, as well as the entire company itself. There was a cool experience because when I joined there were six people. So there were the four co-founders, one who is like doing the designing, and then me and then another intern, who is a graduate master student and from Paris Polytechnic. The cool thing was the four founders, one was Danish, then the guy who wrote the paper was French, another guy who's Dutch and the last co founder is Greek so you got all this multicultural nesting in these EU startups, which is absolutely amazing. They provide a lot of insight into different life experiences as well as the startup environment.
Okay, so veering away from sort of professional academic stuff, what sort of hobbies and interests do you have, what do you do for fun?
It's kind of not as easy now, but games was the thing that I usually play with friends and stuff. That's not really applicable so sometimes move on to more online kind of stuff so that comes out to be either Among Us or more social deduction games kind of stuff. In terms of hobbies, I mean I'm currently doing finance and stuff like that which is kind of cool; you get to learn a lot about economics and current news and stuff like that. I have a lot of alone time now that we're kind of secluded so I'm basically spending more time on YouTube and stuff like that, learning about history and stuff. I mean in high school I used to do fencing, but it's a lot of work in terms of maintenance and all that good stuff so I didn't want to bring it to Berkeley 'cause like doing all the maintenance, as well as like you know just lugging around that huge luggage. I don't have enough time for that, that's how I feel.
I fenced for one year in high school and it was pretty cool, I actually broke a foil, my coach was pretty pissed off.
Oh nice, I mean that's pretty common like in my bag I would usually just have like 10 different spare saber blades that I just kind of switch out of, saber is pretty easy to replace. Well, same with foil but like foil, you have the button and the button is annoying as hell. But yeah I don't know in four years I broke at least ten blades so yeah not fun. Especially in Berkeley I'm not exactly sure about where to get the blades and all that good stuff so I didn't want to deal with the hassle so I didn't continue it.
So we're gonna move into some questions about SAAS, what would you say is your proudest accomplishment in SAAS?
I think that's a weird question because I feel like accomplishment isn't really something that I gained personally, it's more like what I learned over the years right, and I would have to say literally all my machine learning. Like my deep learning foundation was built up by SAAS. My first semester, I was doing the facial recognition project, I learned about convolution neural networks and I was able to play around with it a lot. And this is why I always talk about consulting as the playground for any data scientist, because you get to get down and dirty with the data, and with all these peers that knew what to do, I was able to learn about data, about CNNs really, really quickly and the following semester, I learned about autumn colors and after that I was able to become self sufficient with the knowledge necessary in deep learning. Without taking any of the Stat 150 series classes or 189 or 188 I was able to understand that knowledge without any more instructions, essentially. And after that it also taught me a lot about the managing side of things, after I became a project manager. I don't think of any of these as accomplishments and I don't think of any projects as like certain accomplishments. These are just different problems that I was able to enhance my problem solving process over time, and that I feel like it's not an accomplishment from myself but of the committee itself so teaching people how to work with machine learning, deep learning models. I actively try to teach people about neural networks and stuff like that. You'll see that if you step into the PM role, you'll see that a lot of the Data Consulting members don't come in knowing the deep learning fields, either like computer vision or natural language processing or even the basics like decision trees and like SVMs and so on. You have to take the active role to then lead them to the correct resource, teach them when it gets confusing. And even last semester I spent a huge amount of time teaching people about T-tests and chi-squared tests and KS tests, and these are the things that you will learn a little bit about in AP Statistics. And you'll learn a little bit about it in Stat 135 but then you'll forget about it and in last semester's project, it really brought out how useful these methods could be in a real world setting. I feel like that's kind of the accomplishment of the committee as a whole. There's a certain reason why we do have to teach them, we could technically have a club built up of completely seniors but there would not be a club if we were to do that. Because they'll just leave right afterwards right? That's why it is fundamental for us to have freshmen and a CX, at least in the CX committee being brought up in the SAAS environment, and then they should be getting in Data Consulting hopefully in their sophomore year, at least if they show promise and yeah we do have to pass on the course sometimes and it's hopefully at least a junior you know.
That definitely makes sense, like if you have too many seniors and not enough young students your entire club can collapse in a year. I mean I don't know what's going to happen when Rachel graduates, because she literally runs the club on her own.
Yeah she's really like carrying the IVP side of things. Like she's doing really well. It kind of does make us kind of apprehensive about the future, but a few semesters ago when it was still under SUSA, you didn't see but there were a lot of talented seniors that did go into that advisory position, and there was kind of disconnect so we want to kind of mitigate that as much as possible and. All these interviews and stuff allows us seniors to pass on this knowledge to the rest of the underclassmen so that our club becomes better and better instead of just being replaced by inexperienced people every single year, you know.
Okay, I think that's that's that's good in terms of SAAS questions, we can move on to Berkeley questions, what's your favorite class and professor?
Honestly, well okay favorite class has to be CS61A. If you don't understand 61A don't do CS hands down, and if you do like CS then 61A will be a class you will love. I believe that has to go without any question. If you don't like CS61A everything after that is built on top of it and you won't like anything you do and you'll think everything is a chore. After 61A, of course, 61B because these two things are the fundamentals, if you don't understand or don't like any bit switch majors immediately, please it's for your own mental health good and for your career as well. In terms of the upper divisions, I would say… now this is where it's a little bit hazy because I'm not a big fan of the CS classes as much as the stats classes in terms of upper division classes, so I would recommend 151A, 154, and 153, that would be linear modeling, time series, and machine learning. I would recommend these out of anything else of the upper division, especially in the machine learning field, because first 154 machine learning teaches you about the basics, about how exactly machines are learning from the errors and stuff and how you can improve them. 151A is the basics of all machine learning models, it is linear modeling which is like you'll see everything become reducible to linear modeling systems. Okay, so that's why it's so important, and there are so many problems out there that you could overcomplicate things and just think about it as oh neural networks and immediately slap on some GANs or some auto encoders or transformers, that would be overkill, you could do something as much simpler with linear analysis and stuff like that. And 153 is an application upon linear modeling which deals with the time aspect of things and so that's why I feel like these three are on the top favorite classes as well for the upper division classes. And in terms of teachers, I'd say my favorite teacher would be like John DeNero and Josh Hug because these teachers, they really love what they're teaching and you can see it through their passion, they actively strive for higher and higher quality both in the workload as well as in the lectures, and like these other professors, you want to strive to be if you were to try to go through teaching roles and stuff. Though here's this one thing I do have to mention: there are some really bad teachers, but sometimes you'll think that they're just bad at teaching in general, but actually they're not. One example is when I took Stat 133 which is Intro to R with Gaston Sanchez. It was the weirdest thing, well, it was absolutely horrible But it's actually not because of the teacher. The teacher hates the course and all the students hate the course, nobody likes the course. It was one of the worst things that you could ever imagine. It's like step on through trying to learn these commands and memorize them like why the hell are you memorizing them? You can search it up later on. But then I took 154 with Gaston Sanchez and he was writing a book, he was writing the textbook on the field, and it was absolutely the best class possible and the best teacher to do so. And like I was so surprised because, I was so horrified by his teaching in 133 that I'd already labeled him as a really bad teacher but in actuality it's just not his field; his field is more on the machine learning side and on the theoretical side and he he just knocked out of the park in 154. Similarly for other classes, you might only be seeing one face of a professor and they are sometimes forced to teach those classes so basically some teachers are going to teach some classes and other teachers really dislike teaching that class. Similarly for Hilfinger versus Josh Hug for 61B. Just sometimes these teachers they're forced to teach in general, and they just don't like teaching, you know. You may have some of those professors like that, but usually if they're passionate about what they're researching into then of course they're going to teach well. But yeah I mean sometimes you have someone where every single one of his lectures he kind of breaks us, he's like oh come on guys, this is the easiest question ever but it's rare, it's something pretty rare. But with 61A and 61B, you'll see in the future when you actually have to create classes and actually do object oriented stuff you'll recall like abstractions and stuff just on the fly, you won't even think about it, because it's already ingrained within you, like these two classes teach you a certain mindset. That's not something easy to do and no other classes did it better than just these two classes. No matter what kind of online courses you take, like computer science or AP computer science and stuff like that, none of that. That was all garbage, those all teach you how to use the programming language; that's not important, these two classes teach you how to program. I took the AP CS test and I saw what my my friends had to go through in the class, and it was all just BS so I just read up on it, it was just learn how to Java, so I learned that by myself and just took the test, I don't think it was that hard for you either right, you're a CS major.
OK now some casual stuff: what are your favorite spots on and around campus?
Oh man feels like forever huh. Yeah there were a few spots that are really like, let me just think because after I think after January I started self secluding in our homes and then in March I went back so it's been a year since I came back to New Jersey. I lived on northside, so I live next to SafeWay on northside and my first semester my first year I was in Foothill so most of my places would be on northside. You should probably understand that, because I think most people would go to southside but the reason why, well first thing is the reason why I don't want to be on southside was all the crime, all the stealing, and all the break ins really was a turn off. The way we think about it, there's a chance for you to avoid everything right, but the chance of you like getting hit with that kind of event is exponentially growing higher and higher every time you spend there and once you hit it the loss is extreme, you might even lose your life, which is infinity. So if you do the expected value and stuff like that essentially you get infinity. But like aside from that, I live next to Safeway and there's one store that makes the best pastrami sandwiches, they're called Saul's, they're pretty far up North, if you walk from campus it'll be like 20-30 minutes. It's called like Saul's Delicatessen which isn't really a word that people know how to spell. And close by there's a place called Imperial Tea Court which is really secluded; it has a fun garden area that's like pretty zen-y and they make a pretty good team, so those two are very good places I would go from outside like pretty far up North, and just chill community because it's all the suburbs and there's no dangerous activity outside. It's a white community. Like there's a lot of old people so you'll know how chill it is. There's a lot of coffee shops and tea shops there, so the northside is somewhere I would really recommend for good food, it's called Gourmet Ghetto, but I think they're changing it because ghettos, not a good word to use anymore. But in terms of on campus, oh northside cafes are also really good, I love their pastrami — pastrami is literally my favorite food, by the way, so I always get pastrami sandwiches. Yeah Northside Cafe up on Euclid is pretty good as well, I go there for lunch every single time, especially when you go from Soda it's like going downhill, perfect.
In terms of places to study, I would say… shoot I'm trying to remember, because there is this one place that not a lot of people go to. Wait okay. Oh crap I forgot what it's called, I gotta search it up… oh it's called Haviland Hall. In there there's a library, there are these long tables and more to do but there's almost no one there, there's always air conditioning and it's super quiet.
Yeah I walked by that two days ago.
Yeah you'll always walk by and you'll never enter it but inside is pretty chill. There are some others, I really love to go to these hidden ones or those places that you really feel peaceful because they're not as distracted and you feel like you didn't do work by yourself.
Right at the same time away from your real apartment or whatever.
Yeah exactly, one thing about Moffit: Moffit is pretty loud so I get distracted easily, but Main Stacks - one thing about Main Stacks is that it's too quiet, like you can you hear a pin needle and you become self conscious and then paranoid, and I don't like that. It's just so quiet you feel anxious and then you start looking around. If there's absolutely no one there it'll feel like a horror movie.
Another place that was pretty interesting was Moses Hall. It's right south of South Hall. You go in there, it's like a language place and then I kind of went upstairs and it's pretty secluded but then it turns out that it was a graduate library that I wasn't supposed to sit in. You'll walk by Wheeler and you always walk by Moses Hall, I think it is the language building so yeah technically you're you're not supposed to be in the graduate library but no one checks. But once I was occupying one of those desks that was for one person and then this person walked up to me, was like excuse me, are you a graduate student, I said no, and then they were like, "oh, these are reserved for graduate students only" and there's a big RIP and I was kicked out and she was a graduate student.
Cafe Zeb was the last one, it's in the law building. So the cool thing about that area is that the law building and Haas themselves are paid for by graduate students and they're super rich right, so everything looks pristine. And the great thing about Cafe Zeb is whenever you feel hungry, you can buy one of those breakfast sandwiches and you could see a snapshot as the default image and that looks disgusting but I'll tell you it's fucking great. And the cool thing is it's not completely silent, there's a lot of people, well there were a lot of people there, but yeah you could eat while you're working and you can get more food and the seats are pretty comfortable so you got everything you need; you can stay there for for the whole entire day.
Hilarious, there's a Google review that gave it one star because it's comparing it to Michelin reviewed restaurants in Italy.
Feels bad man, I gave it a four star, I used to do a lot of Google reviews. Yeah I gave it a four star and I said my biscuit was a little oily, but it was really good.
I wonder if I can find you… Found it.
You did?
Yeah all I had to do is search 'biscuit'. Luke Dai, local guide.
Yeah dude i'm a local guide. Used to get a lot of perks but now none of the perks are useful. I forgot, it was like you get some kind of Google Drive space or something like that. Oh, there was one class that I took that is kind of like completely irrelevant now but I'm just looking at the map, so in archaeology building. It's the anthropology library, do you know that there was a museum called the Phoebe A Hearst — you'll see it on Google Maps, the Hearst Museum of Anthropology. I took a class that was Near East Studies 18 or something. It was Intro to Egyptology. Yeah so in there you'll see there's a few art artifacts, basically Phoebe A Hearst was a patron, like she gave money to archaeologists and the guy went to Egypt and stuff and got a lot of artifacts, came back, and this is the museum that has everything and because the graduate student who was teaching, I guess she's one of the lead graduate students or something, she let us go underneath the building and they store coffins, mummies, and artifacts and we got within like a few centimeters in front of the artifacts and actually got to look really close to the artifacts, and these are from Egypt and stuff, it was absolutely amazing. And they're all hidden underneath this building, so I never knew that there was even a museum in Berkeley. Well yeah, that's a place to hit, it's pretty small but all the good stuff are hidden underneath, there's at least 10 coffins, like sarcophagus or wooden so they're super rare, wooden sarcophagi are pretty rare compared to well, I mean there are also stone. Granite's mostly for royalties, but yeah.
Anything else you want to say regarding SAAS in general?
Almost everything I learned about ML, I learned through SAAS.
Is there anything you want to someone who might be reading this?
Do we have a SAAS slogan? What's a good slogan? Shout out to mom; shout out to dad. I love SAAS - it has taught me a lot. SAAS has given me a lot of people I can call close friends and people I can rely on. It's been a place I can grow and hopefully continue to grow.
Anything else you want to say regarding SAAS in general?
Almost everything I learned about ML, I learned through SAAS.