Closer to welcome to the Asos Tech podcast, where we continue to share what it's like to work inside a global online fashion company.

Perfect.

You may have bought some clothes from us, but have you ever wondered what happens behind the screens? Hi. I'm Carmen Fletcher, and my pronouns are she, her, and I'm Associate QA. At Asos today's episode, we have the wonderful Eleanor Low with us to talk us through a day of machine learning at Asos.

I'm Eleanor. She her. I am a lead machine learning scientist at Asos, working with an Asos AI, and I specifically lead the AI pricing team.

We're going to start with an Icebreaker. What piece of tech would you say you can't live without?

A very wary starting flame wars this early in the podcast that I actually think the one piece of tech that I have been using since grad school is Sublime. So I don't know whether you guys are fans of Sublime. I use it for literally everything coding, but also for taking lots of notes. It's just got so many great little plugins and whatnot. So my entire brain is organized in Sublime. Definitely couldn't live without it.

Nice. Okay. What did you want to be when you were older?

When I was older, I actually wanted to be an artist. It's very far away from what I do now, but I used to be a humanities nerd and a literature nerd. I used to want to be a dancer, a lot of stuff like that. And now I do math for a living, so I don't know, something's gone wrong in between, but I guess I can't really complain. I do love my job.

I think tech has a lot of creativity behind it.

That's true. Very true.

Because you can do so much with it. And I think having a creative brain does lead towards all the technical stuff as well, and you get to look at things in a different way.

Yeah, for sure. You get to build stuff for one.

Yeah, that's every creative stream building everything 100%. So how did you go from that to getting into tech at Asos?

Tech at Asos. So I've been working in AI for a while before Asos, so maybe I'll say that first. The entry to AI for machine learning people is often via lots and lots of education. In my case, I did a PhD. I worked as an academic for a while all over the place. And I guess while I was doing that, I got really good at some bits of statistics and modeling and math and stuff like that and ended up here at some point after working at Delivery and at Facebook on the Asos bit, actually. I think as a young woman living in the world, obviously I knew Asos, obviously. And one thing that's quite funny is that when I got recruited for Asos this time, I actually pulled up an email at some point, in fact, years and years ago when I was trying to get my first machine learning job. And this is actually an email from me just proactively reaching out to whoever was the recruiter back then saying, hi, machine learning experience. Are you guys looking for juniors and stuff like that? As a young woman living in the world, asos has been part of my life's journey. So for me coming to work here, it was really cool to come and obviously see behind the curtain and participate in something that is obviously like a great product in the world.

How long have you been here at Asos?

I've been at Asos for, I guess about two and a half years now. I joined Smec in the middle of the pandemic when everything was on Zoom.

Was it quite tricky joining during a pandemic?

I think I am so fortunate to have joined the team that I joined because obviously joining any group of people fully remote on Zoom can't leave your house. Obviously that sounds awful, but I had a lot of good fortune to land at a team that is just super fun. Everyone is a really nice person, number one, which it sounds like a nothing burger of a statement, but it's really not. One really cannot assume that, I think in different kinds of work environments, but also everyone is just really chatty, really comfortable being themselves. There were like a couple of times where I remember my first week, one or two people would turn up at stand up with your hair up in a big towel. And the thing is, they're coming to stand up like that, but they're also super audit. They're properly doing their job, they know everything. So it's like bam, bam, bam. It's a really nice combination of knowing you're going to be able to do good work and the bit of you that wants to work and wants to do the thing that you're interested to do. That part of you is going to find a home in this place. But also, people are cool. There's no need for stiffness or whatever. There's a culture there for you to participate in. That's what I found when I joined my team, and it still is the case now that it's like that, but even more so, I guess, now that we can all go out together. I feel like I was very fortunate to land it in that environment, especially in the middle of a pandemic I've.

Been in on a couple of days where you can really sense it in the office. You can really feel that everybody's just really so full and everyone really gets on and you can see how everybody goes and pairs and stuff.

Yeah, there's never been any sense that you can't bring yourself to work, you.

Know what I mean?

And that's not to say that you're not expected to do good work and people are going to call you out if you're not behaving well, there's that kind of honesty that comes from a safe space as well. But yeah, it's a safe space to be yourself.

I know very little about machine learning. Can you tell us a little bit more about what you do on your day? Safe job?

Okay, this is going to be hard, particularly because I lead a team, but I'm going to start right from the beginning. So we're AI team, we work on AI products. We design them, we build them, we own them after they're built. But we talk a lot to the operational and retail bits of the business to try and figure out what the problems are that they need to solve, to run the business, to scale the business, all that kind of stuff. So we do a lot of work to figure out what the domain problem looks like and then we go in. We use the knowledge within our discipline to build products that are going to scale. High quality decision making. Obviously there's the recommender. Lots of people might have heard about Rexys and whatnot I specifically work in the pricing area. So a lot of the stuff that we build is around how do we forecast demand given different kinds of price points, like how do we design algorithms that are going to enable the company to operate efficiently. So in terms of the business side of things, that's what's going on in our heads all the time. We are classic software engineering, cross functional team. So our team has scientists, we have engineers, we've got a product person. Some people have Bas as well and different kinds of scientists as well. But everyone's kind of working together and collaborating to do that end task. In terms of what I do specifically in my kind of day to day, at heart, I'm a scientist, so I do a lot of thinking about how scientific methods in AI and machine learning can be used to solve product problems. There's a lot of very technical thinking that I do. I work with the scientists and my team to progress projects. I work with the engineers to make sure that we're building what we need to build to make sure that the systems that we build are resilient, they're maintainable all that kind of good stuff that we balance, how we're thinking about tech debt and long term investments. And I also work really closely with a product person who she's the main interface that my team has towards the business. And she's like the voice of product on our team who makes sure that we're always focusing on the most impactful thing, that we're always in tune with the business and stuff like that. That's all the kind of stuff that I do over a long span of time. The other unique thing that we do, I think, at Asos is that we write papers. So how that works is often as we're trying to figure out what needs to happen in order to solve a specific domain problem. We do end up reading a lot of research and doing a lot of technical thinking to figure out how to solve the problem. We have a prototype. We build a solution, we'll a B test it. We'll get a rigorous sense of whether what we built actually solved the problem or meaningfully helps the business or not. We'll ship it, and then after that, we'll write a paper. And the papers go into machine learning conferences, which there are a few conferences that are just attended by lots of people who work at AI and Tech, and they're a really good forum for us to go there and obviously show off our work a little bit, get some feedback from other people about what are good ideas, look at what other people are doing, and just cross pollinate ideas with people outside of our specific company. So that's a really big part of what AI does.

What's your best day at Asos like, going through what little meetings you've got? Let's say maybe when you're in the office, because you probably have social things that you guys do as a team as well.

I actually had a pretty good day yesterday. There's one part of it that I think a lot of people who start as ICS are technical people at heart, but are responsible for running a thing more broadly. I think anybody in that position can relate to what I'm trying to say. Yesterday was our in office day. Obviously, we got to go in and see everybody have a bunch of conversations, including with our department's leadership, around some things that they wanted to happen. One of the conversations that stands out to me is one of our managers is working really hard to try and set up community structures to take care of all the scientists. They want it to be the case that people feel like they belong, that they're comfortable and whatnot. And so we had this meeting talking about how we can use some of the existing structures to achieve those aims. And what I really like about that meeting is that the manager who was running this and who is soliciting feedback, he really cares. He's a manager who really cares that this is something that does happen. And the amount of thoughts and reflection that we did about how to make this good, how to make sure it's like serving the purpose of making everybody feel at home and psychologically safe and able to collaborate and all that kind of stuff, I think it's things like that help us maintain good culture. And so it was part of a nice day for me because I like being in a company that cares about something like that.

Did you jump straight in as a lead at Asos, or where did you begin? How have you moved up to becoming lead?

I didn't join as a lead when I first joined Asos. I really got to just focus on my discipline, which at heart, I'm a technical person. And it was really enjoyable to just get to do that and execute on my discipline. And at least within Asos AI, there's just a lot to do and there's opportunity in that. And so as people move on, people move to different areas over time. I got asked to lead this team. Part of that is some amount of people management. As the person who leads the team, I'm responsible for motivating people, lining them, putting in mechanisms to make sure everyone's rowing in the same direction, and making sure that there's a lot of success for people to participate in. It's not a role that I necessarily would have sought out had you asked me on my first day at Asos. I think, like a lot of technical people, sometimes I think I just want somebody to dig me a hole and put me in the hole and shy.

Away from all of the people.

Yeah, because you never think to do it. Because there's this other thing that you really, really enjoy. To be honest, I didn't really necessarily expect that I would enjoy that kind of thing that much, but I was put in that position.

It's nice that they've helped you grow in that way and sort of like lead you into a different path that you kind of didn't see that you were going to go down.

Yeah, I guess so. I want to be honest and say that it's like a lot of things that happen, I guess, in everybody's career are not necessarily deliver it. It's not as if it's not quite so top down and organized in terms of, oh, let's create this thing. And that's what I mean by there's a lot to do and that there's opportunity in that. In my case, it really worked out for me because I was pushed to do something that I wouldn't necessarily have sought out.

What advice would you give to someone who wanted to either work in machine learning with AI or someone who wanted to come into Asos Tech?

I think with AI specifically, it's really important to know your stuff well, to focus on the fundamentals and to understand how the technical area works. I think this is a particular thing about AI. At the end of the day, if you're going to be responsible for AI products, it's your problem if you launch it and it doesn't work. So you really need to properly understand what it is you're building. Have the ability to build it in a way that's resilient. That would be my advice. And actually, beyond that, I would say we're a place that also allows people to learn. I think a lot of people can find AI very intimidating because it's so complicated. There's lots of people with PhDs and whatnot. All of that's true. If there are lots of people I can think of where their interest is genuine they're willing to work hard. Yes, there is a lot to know and it is difficult, but we're really respectful of that and there's a lot to build, so there's a lot of opportunities. Learn while doing and put your skills directly into action. So, yeah, come join us is what I would say.

AI and machine learning being a part of your daily life through work, and it's obviously something that you're very passionate about. There are things that you do to incorporate into your home life that goes with all of this, any AI that you use.

So there are two things I'm going to do. I'm going to tell you one example of one example of an algorithm I love, and one example of an algorithm that I don't know. Let's start with the good one, which is I really love to play with Spotify's algorithm, so I really love music. Actually, in a previous life, I came very close to doing a PhD in music perception, and Spotify just has a great recommender. I think if anyone's interested in playing with AI systems in their life, just go and interact with the product. Give it signal about what you like, what you don't like, that kind of thing. But also with Spotify, one thing that I also do a lot is I'll just seed a playlist with stuff I like. Ask it to turn on the radio based on particular playlists. It's really great at doing that. And actually, if you vary things up, sometimes you're like, oh, I feel like listening to jazz. You got a jazz playlist. You do that. But sometimes you want to do like a mashup of a bunch of styles that you wouldn't think to put together. You put them as you go, you see them, and you can see what the brain thinks. It's a really great tool for me for discovering new music and things like that. So that's a good example. I recently got a Google Pixel phone. Obviously, Google and AI, like, everybody knows they're great. This is one of those areas where I think progress in AI at some times can be a little bit slower than what the hype cycle might suggest. So if anyone has a Google phone, obviously you can talk to it, ask it to do things, oh, set me a timer. What's the weather going to be like? But then when you start doing that, you get really excited and then you start trying over time to talk to it more again, it's your more complicated things and it just can't handle it. And it's such a bummer. I can't wait till all this technology needs to evolve gets to the point where you can really use it really powerfully in your life. Join us next time for more stories and insights from behind the screen at Asos Tech.