Bay Area AI: Alyssa Wisdom Interview
Recording: Bay Area AI: Alyssa Wisdom Interview
Hello everybody, I'm Alexey Krabarov, the organizer of Bayer AI and it's a meetup here at Metis San Francisco, which is a data science educational institution nationwide. And it's a join meetup with H2O, San Francisco Big Data Science meetup and Metis educational meetup. So we have excellent program tonight, we have two talks, one talk from Antonia Villarobanos and one talk with Alyssa Wisdom from Square. Welcome Alyssa. Thank you. So Square is a very well known company, right? Everybody uses it. Tell us a little bit what you do at Square, how you become interested in the topics space. Yeah, so currently I'm a product analyst at Square and what does that even mean, right? So how we do product analytics at Square is that we embed data scientists into each of our products
And we use data science in a mix of analytic tools to help inform the product on how to move, why to move and where to move. So yeah, that's, I really, I've been working there for about over a year and I really love the company. I love the mission of economic empowerment, what we do. And I also love the data culture and how we use data. I know that you asked how I got interested in the topic modeling space, but honestly the way that it is structured at Square is that anything that interests you, you can, you can either learn or you can show, share your knowledge in. And it's, it's a really great place with a lot of amazing data scientists. So we do any and everything there. We're not necessarily limited to topic modeling or we're not limited to like one type of machine learning
We do anything that the product needs to succeed. So it's been a great learning environment so far. You know, I was always wondering, right? Because like, you know, every time I swipe my card, every time I do something, I'm thinking like all this data. Yeah, so much. Like there's data somewhere. Yeah. And we have a team of data scientists just like looking into it and trying to figure out what does this data mean and how can we use it to better improve our product and really just, yeah, help the customer as much as we can. So, you know, I like, you know, I don't ask you to reveal anything, but here's what I thought immediately when I started using Square, right? I think it should have changed behavior on a giant scale because basically it offers you to tip, right? Okay
And offers you to tip one, two, three bucks. Basically the operators can set fixed scale, right? And so most coffee shops, like I drink a lot of coffee. I can sustain myself on coffee alone. And so they have this 15% set up for tips, right? And so sometimes I buy a bag of coffee, it's just 20 bucks, right? And it kind of, the default will just use that, right? And so, kind of normally in the old economy, right, you would have to use that. You would tip with change. So it will not exceed a dollar. But now, like, you basically, like, you're forced to say, okay, I'm not going to pay you three or four bucks because I'm paying a bag of coffee. I'm going to use custom and I'm going to give you a dollar
But it should lead to enormous amplification of tips. Do you see, do you do anything with that? Like, do you see, like, a lot of tips coming into the merchants? And unfortunately, I don't work with payments data. And even if I did, I'm not quite sure I can share all of that information. But I will say that in all of our products, including our major, you know, process, payments processing, important sales system, we make the product so that it induces a type of, or that it changes the entire experience of, you know, payments and then how a merchant takes payments and interacts with the customer. And so I think in everything that we do, we're really trying to drive that home. Yeah, I mean, I'm not a big fan because I see Square Terminal and I know I can use my carry card, right? So it's like suddenly I come to some festival and they use cash. I'm like, what is this? Like, can you guys set up a square? Right? So I'm a huge fan. Yeah
And it knows all your carry cards, right? Basically, it sends you emails. So it's kind of, so you guys can track a lot of behavior this way. So where do topic modeling, where does it come handy in the Square ecosystem? Yeah, no, and that's a really great question. And that's something that I'll be talking about today. One of the reasons why I actually started looking at the topic modeling personally is when we wanted to analyze churn reasons, right? So some companies will give you set categories of like, why are you leaving the company? Please select from here so that we can understand that. And normally it's based in like some type of intuition, probably past customer feedback. And it's really easy to analyze because, you know, you have set categories and you can see X amount of people chose this, X amount of people chose that. But sometimes it can be really limiting and doesn't really evolve with the customer very well
But when you have like free text fields, you get a rich amount of information. But it's a bit more difficult for say like our product managers or just, you know, the regular person that parsed through that. So that's where topic modeling comes in. And that's where I actually started really diving into it and using it as a tool to help provide that data solution to my stakeholders. Yeah, I like to bring people together. Our second talk is about NLP as a middle way, right? And so if you need a lot of parsing, you know, I think Andrew can help with this. Awesome. So interesting, interesting
And so tell us a little bit about the data science culture. How does it connect with engineering, right? Like in these companies, usually there's this dichotomy like you're an engineer, you're picking on some code, or you're a scientist, but you have to work together and it's different in every place. How does it work at Square? Well, I can only speak for my experience. I can't necessarily speak on behalf of the company, but for Alyssa Wisdom at Square, it's been awesome so far. I really love it. As a product analyst, I'm embedded into the product. So I work very closely. It's a team of me, myself, my product manager, a bunch of engineers, and we work really closely together to make sure that one, the data that's coming in looks the way, is coming in a format that I can do something with it to help inform them of how they should keep engineering or improving the product
And so even today, I probably have like maybe three meetings with my engineers and my team, and we work very closely together. So that dichotomy does not exist for me and my function. Yeah. Well, this is great. It sounds like a good culture. Yeah, it's a great culture. And maybe, you know, one last question I'll have is like, what is the most surprising and unexpected finding you saw when you looked at all the data? All of the data? That is a very . I look at all of the data every day, and every day I'm surprised and excited
And that's very broad. But, you know, I think one thing that really surprises me every day that I go to work is just that by looking at the data and seeing all the hard work that we've been doing, just the different ways that we use data at Square to economically empower our merchants. And I think that surprises and excites me to go to work every day. So this is awesome. You know, thanks for coming. We're looking forward to your talk. Yeah, thank you. Thanks.