In this episode, I talk to Diana Mounter, the Director of Design Infrastructure at GitHub. Diana traveled the world and lived in many different countries – even continents. She started as a print designer and spent some time in government before she got into web and design. Now, she leads the design systems at GitHub.
We talk about:
- what design systems are and why we need them,
- how GitHub deals with legacy code and refactoring.
- how the designer role interplays with other roles at GitHub,
- how and why designers do code reviews,
- and how GitHub strives for inclusive designs that make everyone feel like an expert.
Michaela: [00:00:00] Hello, and welcome to the software engineering unlocked podcast. I’m your host, dr. McKayla. And today I have the pleasure to talk to Diana Mounter.
But before I start, I wanted to update you on what’s happening in my life. This is the, I have the last one code review workshop for this year, where I still have a few spots left. So if you want to make code reviews is super power. Now is the time I will resume the workshops in January prior to break, right? Yeah. The reason for that is that I’m finally working on the code review tool. I always wanted to work on, remember when I said I want to do it and there’s just no time. Well, I made time for it. The tool is a code review analytics tool currently focusing on GitHub data and it helps engineering teams too. Okay. Understand their strengths and bottlenecks for the code review practice. Right now, it is in the MVP state, but I’m already working with one rate company that wants to understand the different cultural practices followed by various team and organizations. There’s no official side or anything, but I’m interested in the code review analytics tool. Send me an email at email@example.com. I will link my email in the show notes, but now let’s get back to Deanna. Yeah. Anna is the director of design infrastructure at the tap Vienna travels, divert and lift in many different countries, even continents. She started as a print designer and spend some time in the government before she got into bed. And design now she leads the design systems at guitar. Well, look, I’m Deanna on my show. I’m really thrilled that you’re here.
Diana: [00:01:41] Thank you for having me.
Michaela: [00:01:43] Yeah, sure. Well, yeah, I’ve never talked to their director of design infrastructure. Can you tell me a little bit, what’s your day to day work look like and what are your responsibilities?
Diana: [00:01:56] Yeah, so my day to day is, you know, it’s a, it’s a fair amount of meetings, but I would say that. A lot of the work that I do is T is like to help shape projects, help develop roadmaps and strategy, and also people management. So I think one of the things that we do at get hobbies in the design org is that design leads tend to also be like experienced in, in what that might seize do. So they have a lot of input into the design, but also people manages. Would you like me to tap into what design infrastructure is?
Michaela: [00:02:32] Yes, please.
Diana: [00:02:34] Yes. Yeah. So I, it might not be the most commonly used like a title, I guess. So design infrastructure, I think is. A good way for me to describe a bit more about what we do. I think design systems are a big part of it. I would almost say that design systems are sort of an output of, of what the design infrastructure team does. So we have a get hub. It might be different a bit in different places, but it get hub. A team has like kind of two for, I guess, areas. One is the design system itself. And the other is how the design system is used in the core experience of get hub. So we do a lot of work on the actual sort of like infrastructure of, of GitHub in terms of. The actual like design and the user experience of get hub.com. Another way I feel like is, is maybe a tangible way to sort of describe designing infrastructure is actually to sort of comparison. It’s a sort of real world in the real world, they just described like Amad structure and in terms of hard infrastructure and stuff. And it’s very much sort of works for like what we do too. So our, like how the infrastructure, like UI components and we have a couple of different implementations of those it’s tooling, such as like LinkedIn’s unit tests, that sort of thing. And then like release distribution. Like how do we like distribute the actual components to different websites as well as github.com and then the sort of stuffed infrastructure is, is the citizens. Processes and stuff that we have to do to maintain the health of, of the design system. So that will be things like we have, like a first responder rotation, office hours, a contributor group, onboarding processes, reviews, that type of thing.
Michaela: [00:04:24] Okay. And so if I’m thinking I’m a software engineer now at GitHub, how do I connect to design infrastructure? So I’m actually getting UI components from you that I can then reuse or just, you know, you sort of out of the box. That’s how I would imagine it, but I’m not sure if, if that’s the correct. Understanding.
Diana: [00:04:46] Yeah. That that’s the happy path for sure. Like that’s what we hope the experience will be. So I think, you know, I’m working on the front end. Yeah. Get herb is likely to interact with the design system, which is called primer on github.com. That’s probably the two main ways that people would interact with that system. Um, in terms of like the actual code implementation. One is his primacy assess, which is really where we started with the design system. And the other way is through our component architecture in get hub.com, which is called few components. So that’s a rails implementation of components, similar to what you might use with in like a react implementation. And so at the moment, yeah, the reason why you might sort of interact with both is because we’re in a sort of like transition. Phase of moving to towards components, but we’ve still got a lot of work to do. Another way that engineers would interact with the system is they’re building in react. We used to have a react component implementation, also a documentation. So how to like guidelines on how to implement and, and then, yeah, probably code review in NPR. It’s like we review a lot of this sort of. Sort of PRS that tension UI implementation, basically.
Michaela: [00:06:11] Okay. And so from, from an organizational point of view, the hierarchy and how divisions are located, is it that, you know, you have like design infrastructure is their own thing where, you know, I imagine a lot of designers and, you know, sit and create those infrastructures and processes. And on the other hand, you have like the engineering teams and then. The engineering teams just interact with the infrastructure or is it more that in the engineering teams, they are also designers and they connect to the, to the design infrastructure team.
Diana: [00:06:45] Yeah, that’s it. That’s a great question. So on my team, I have both designers and engineers, and so the engineers tend to work on the, the actual UI components in our different implementations on like things like the tooling, like lenses and unit tests, like. I mentioned earlier, they, they managing cut their releases. It’s the engineers that are more often working in. On the actual github.com products, then they’ll also update the different versions of, of our primary packages and like test that and make sure that’s working. We do a lot of updates.com itself, so like a lot of refactoring and bringing the front end and the design forward. Design our team is like pot is the product design. So yeah, we have like a ton of designers on the team T summit. Some of them are more technical. Some of them are working more on the visual side of things. And, um, so we’re part of the product design team and product designers tend to work in. And squats with PMs and, and engineers in the engineering org. So it’s kind of like a sort of a matrix set up, really. So we’re our own org. We also work with a lot of other product teams, basically.
Michaela: [00:08:06] Very cool. And so, well, one of my favorite topics called reviews. So for example, let’s imagine we have to improve the code review experience for who and how. Would designers be part of that. Right. So, yeah. So would it be like one product team that would say, well, let’s, you know, redesigned, for example, that experience on that one page, let’s say per request page. Right. So, um, and we go through data and then you have some designers, as you said, in the team, in a squad that would redesign really how it looks like people would implement it. Like the engineers would implement it. And then the buttons, the components we will really pull out from infrastructure. Is that right? Is that how I would imagine it, or how would that work?
Diana: [00:08:54] That’s, I’m fairly close to, I think how we work, I’ll add a little bit more fidelity to that. So, I mean, I think that it’s often easy to think of design is sort of, you know, pushing pixels, but really designers are thinking about the sort of end to end flow. What’s the customer experience also in working with PM and engineering. Leadership to, to think about like the business sort of impact as well. And some of, some of our designers work on things that, you know, don’t really have that UI. So we have design is working on and her boss cl GitHub CLI for example, it used to be called hub. So I, I was using the outdated name there. But yeah, we have get hub CLI, which is like, Oh, tech spec based, but it’s still like stuff to design there. It’s still yeah. Experience. You’re still sort of thinking about the flows, the user experience and, and interior, like how that interfaces. So I think, you know, and designers have a big role to play in that overall experience, working alongside, um, engineering and product. When they are, they’re often like using the system, but also influencing the system. So yes, you, you, you mentioned like, will they grab the like buttons and stuff? We, we have a lot of content and components in our system. And I give her a lot of the sort of foundational way, but we also have an, you know, honestly still working on building up, like how will these pieces connect together? In fact, we actually just did a six or seven weeks of a, like a whole org sort of, I, I want to say hat quake, but it was, it was several weeks. So it was more like a hack month, really, of all, a lot of, most of product design and design system. And engineers like working on some of our, like UX debt together. And so product designers, you need that using the system a lot of the time, but they’re also breaking it. They’re finding the places where it doesn’t cover and you pattern that they need for their feature. And so there, we really want to work really closely with product designers so that we get that feedback about what, what the system needs to cover, how to evolve it. And so it’s really great to have those contributions from, from product designers and engineers working outside the team as well. So. Yeah,
Michaela: [00:11:24] and it sounds really amazing. One question has popped in my head when you were talking about, you know, having all these engineers, you have the product designers, maybe product managers, you have a lot of internal. People that are also your customers, right. You could have is a product that most of the engineers or all engineers and most of the employees of GitHub will even work with, right. That they will experience that they will use. So somehow you’re your own. Uh, you sandwich is very handy because you can, can use that knowledge to evolve the product in the right direction, but you also get her information from outside. And how do you do that? Like, do you do user studies, do you ask really larger? Do you have studies with larger numbers participants? They are, that are coming from outside that you ask, you know, how to improve the system, or how does that work? How does that process work?
Diana: [00:12:19] Yeah. Yeah. So you, right. It’s kind of great to like, be able to use, get hub, to build, get hub. Our team is like, Kind of particularly fortunate because of the fact, then we also married an open source design system. We get to use a lot of the features of get hub, but also like experience it in a way similar to some of our customers. But it’s also important to realize that when the way that get hubby’s is get him, it’s not necessarily like how every customer uses get hub. And so it’s, we have to be careful to not fall into that trap that. If it works for us, that it works for everybody. That’s definitely not true all the time. And so, yeah, it’s really important for us to talk to customers to understand that. And so, yes, we have a customer research team. We also, which is like, you know, if we want to do usability testing interviews, the studies, those are the people that we will work with. A lot of PMs and designers do some, some of that interviewing and, and, and sort of learning from customers that themselves. We also have other ways to interact with our customers and get feedback. So we have a feature preview feature, which allows us to, to make available certain features to customers and they can opt in to try them out. And that’s something our team has definitely made use of in the past and with some of our like big overall. So it was really a really great way. To get feedback week. Cause then we can look at both data, like look at how the metrics are tracking. Um, like for example, I really basic signalize is people that are opted in like they Oh doubt. Right. Is really, and that’s a signal that it’s not working so well, but it also allows us to get like direct, written feedback from customers. We also have a, like a maintain is great that we can have like discussions with. And then sometimes we’ll like, If there’s particular individuals or groups or companies you want to talk to, then we might reach out more directly to them.
Michaela: [00:14:22] Yeah. One of the things that you. Talk or passionate about is inclusive design. So what is exactly inclusive design and, well, I think a little bit, it plays into what you just talked about. So how, how, how do you make sure at GitHub that, you know, your design doesn’t leave out large parts or even small parts of, of, you know, the, your user base.
Diana: [00:14:46] Yeah, good question. I think that’s a good question. Lots of times though, because they were all kinda questions. I think the best way that I can think of to describe why I like the framing of inclusive design, is it, is it just helps you think about. I think, how do we include everybody into this experience and to get help, wants to be, you know, a home for all developers? How, how are we going to do that? If we’re not trying to think about, like, how do we make this an inclusive and accessible system for everyone to use? Oh, a lot of the time we talk about accessibility, which is incredibly important. And I see that as a sort of subsection of, of inclusive design, but I think it’s easy to sort of quickly sort of think about compliance and boxes that you need to check off rather than sort of sort of thinking how can I make this a great experience for anyone? And so. Yeah, inclusive design is I think, honestly, something that we have to continually work at a can be how we offer that experience can be influenced by many things. One of the things that we really sort of tangible, sort of simple examples, even if it’s not simple to implement is thinking about Kolomiets. If you’ve been on the Twitters or I’ve seen some of my other talks, like we know that get, get her abusers would really, really like dark mode, for example. And there are accessibility the reasons for why people might want that. And then that’s just like, this is a preferred experience to me. And so thinking about like, how do we offer something like doc mites and like other sort of color modes or, or like, Choices for how a user experiences yeah. Get help. Like the actual sort of UI design is a way that we can is one way that we can make that experience more inclusive and to people is incredibly hard and it gets as good and harder and harder since they’ve been at get hub to build an experience that works for everybody. We’ve got open source maintainers. We’ve got people. Learning to code. We’ve got an enterprises, we’ve got people using it, sort of, you know, dead, it’s a utility, it’s a critical part of their workflow. And then people that are using it to sort of, you know, for creativity, they’ve got an idea and they’re sort of, sort of starting yeah. A project on get hub to sort of explore that idea and. And test it out and maybe share with if he friends or coworkers. And so I think that we increasingly have to think about how do we, where do we, where is it important for us to give customers the more about the actual, like sort of core experiences get hub? And I say, I think inclusive design is a really important part of that.
Michaela: [00:17:36] Yeah. Somehow I can see how, if we start thinking about that we come into this or we could fall into this trap of, you know, a hundred different decides, right? So one, one person is a power user and they want all the possibilities right there. And, and use every feature that could have has available. And then we have maybe somebody that’s occasionally on GitHub. And so, uh, what I hear very often, and I’m, I’m not very deep into design, but is that you have a very simple or very simplistic design at the beginning. It should be very simple. There shouldn’t be a lot of options. And then people start to dig deeper. I’m not sure if that’s really the best, the best advice. How are you approaching that? What’s your philosophy behind that, right? Like if you want to have, yeah. Novice people work with GitHub and also advanced people have all the power with the features that are available, but without, you know, overwhelming somebody that isn’t very often on GitHub or not using everything.
Diana: [00:18:39] Yeah, it, I mean, I’m not, I don’t think that there’s like a three step process to sort of figuring that out. It’s, it’s a thing that we, that is challenging and that we have to keep working at. I think the, some of the things that we discuss and, and, and try and think about is how can we make everyone feel like experts and make everyone feel like power. It uses, I couldn’t give you like, here’s the X XYZ of how you. Exactly do that. Cause it’s very dependent on like each feature, but that’s a conversation we have. It’s like, how can we make everyone feel like experts? We really don’t want people to feel like that, you know, they’re not smart enough to use, like get hubs UI because they can’t find where things are. And I think, you know, we’ve got, we have scaled the product and we’ve added more features and we have made that experience more difficult in places than it should be. So that’s definitely something that we. We are working on right now.
Michaela: [00:19:37] I really like this idea of making everybody feel like an expert, because there are so many different ways. These are so many questions. I met so many goals that you could have, right. Making a person feel happy or, you know, satisfied or, but. Having them feel like an expert and really like that independent of their experience level. I think this is something that can be, you know, you know, stuff or designing your system. And the question that you could ask yourself. I mean, if you achieve it, this is another question, but I think it’s a very good, it’s a very good goal to have a really like that.
Diana: [00:20:08] Yeah. Yeah. And I think it’s like thinking about not having anything, be a barrier to like core interactions and like having those power features that think that is a bonus almost to the experience. Like it’s an add on it’s it’s not necessarily. You have to spend like 10 hours learning how to navigate to a pull request in I’ll just use, get hub. It’s like, we want not based sort of cool workplace to be very obvious and easy to follow, but then as people like, you know, if you’re using GitHub day in, day out and building that muscle memory, how can we like, you know, I guess. Find the opportunities there for the people that are using all the time to sort of like build on top of that experience. Like make it quick. I have to decent things like am let them customize certain things in a way that like enables them to sort of really mold get her to, to that flow. Yeah. Like the, the, sorry, I was just thinking like, There’s a, it’s like that you’re a sort of keyboard, right? Your computer keyboard. Like you want like people to be able to, you know, the letters are in and type it, but it’s customizations you can do on top of that to really. So that around you experience, maybe you want to, I’m in a more comfortable position that you want different sort of like types of keyboard keys, because you, that works better for the, for the way that you are using your computer. If you’re typing a lot, you want some delight in there. So you want some colors because will make you feel happy, but maybe you’re like mapping different keys to operations that you want to do, like really frequently. I think those are the sort of like, Things that you want to make an add on and not get in the way of the core experience. If that makes sense.
Michaela: [00:22:00] Yeah. Yeah, definitely. And I mean, one of the things that came to my mind when you, he said, well, make everybody feel like an expert, and this is really a tiny thing, but it’s one of the first things that if you go to GitHub and you want to clone a project, you have to get bash command right there. Every time you’re clicking on that. You have it right there and it’s somehow not intrusive, right? It’s not like, Oh, you think I have seen that hundreds of time? How can, it’s more, it’s subtle, but it’s there. You don’t have to look it up. I mean, even if you’re cloning a system quite often, if I do it, I don’t know, second month I might look at it. Right. I might look it up just to be sure that I’m doing the right thing. Uh, but here I know it’s, it’s there. I don’t have to go anywhere. Right. So there’s a little bit reminder. Yeah. So this is one thing that came to my mind. Well, we talked a lot about design systems already without really introducing it. And yeah, I have a lot of my listeners and myself, we are engineers. Uh, so can you explain it a little bit? What are design systems and you know, how do we use them? Why do we need them? And yeah.
Diana: [00:23:10] Yeah. Um, gay question again. Um, so I would type that like a lot of engineers who are working on the front end interact with design systems or have, have some awareness, but basically a design system is, you know, essentially. Thanks sense of sort of rules and, and principles like codifying code in design. So like the implementation of our system exists in a few different forms, right? So it’s it’s components. And for us, we have, you know, a few different implementations. It’s also on the design side. It’s also those components existing in designing tools for designers to work with. Yeah. And then the connector is really the, the sort of API of the system, right? It’s the bridge, it’s a design language that you’re hopefully developing to make it easier for designers and engineers to have conversations about. And the UI. So that they can say, I want a three column grid and you know what that looks like, where I want the primary button. And they, they knew what that is in both design and code. So I, uh, I like I would go is really to provide that sort of cohesive design language so that people can represent the brand and the product UI in a consistent way. But it’s also about efficiency. So we see, you know, in the past we’ve seen people reinventing the wheel in design and code. Like we don’t want someone. The really simple example is like, we don’t want. Yeah. Someone to, to waste time, we redesigning our primary button and recoding it. Right. But that’s, we want them to be thinking about the feature and the, the, the flow and the, the core experience. So whenever we can find that we can solve like a problem once that can be reused many times, that’s like something that might end up in the, in the design system. I think like over time, we, with those sort of very basic building blocks that those are now we’re sort of thinking. At a bigger scale, like how do you, like many, the components fit together? What are the sort of longer form like interaction patterns and like what sort of like onboarding experience, for example. And so it really, yeah, there’s a, you know, systems are kind of like, Assistant a system of systems. There’s like many parts to it. There’s a rambling answer. But did that, did that cover it?
Michaela: [00:25:44] Yeah. Yeah. And well, especially from the last part, what I got is that, so how you started having all these little individual units, right? Each button is a unit. Each input field is a unit, and then you say, well, it’s actually not only the input. It’s how it looks. And, and you package it and so somebody can take it and use it somewhere else. And it’s the same, it’s the same thing. And then you say, well, not only the input in the button is something, but a beach is something. So somehow you have like this little system, then you have a larger system, which is a, is it. It’s a page and then you have the system of systems as you’re sad Valley, it’s not a page, but it’s a page flow. Right. So it’s, that’s what I understood when you say, well, how is the onboarding experience, right. So you’re thinking really about when do we want to have the user, uh, see message, for example, do we have, you do have to click three buttons or, you know, goes through three pages or something like that. And so, so it gets more and more. On one hand complex, but on the other hand, a little bit more mature. Is that, is that what you experienced at GetApp over the time?
Diana: [00:26:50] Yeah, I think it’s largely to do with scale. So saying, yeah, it, you know, you have to start somewhere, so hopefully it does get more mature you as you grey the system, but as get hub has. Introduce more features as the team sizes grind. We’ve had scale the system with it. We started off with a very we’ll make like utility first, a very functional sort of system. That was lots of small building blocks with a lot of flexibility. And now we need to sort of have more constraints and sort of larger pieces of the system sort of plugged together and a lot less variation. So that means we’re sort of choosing to reduce some flexibility in places where the, the, the value of flexibility doesn’t really pay off, if that makes sense.
Michaela: [00:27:41] And so. One of my question that comes to my mind. If you’re talking about something like that is how does technical debt of for example, legacy code, right? How does that, is that something that’s also a problem for design systems for designers that you say, well, you know, now we want to have a fresh logo. We, we rethought what people actually want or how, you know, how to the UI should look like, but now we have this system and it’s in so many places. How do you. How do you change that over time? How do we evolve a design system when you work with older parts of the system?
Diana: [00:28:18] Yeah, a lot of refactoring, but I think, I think like, like at the point of a system is to be built from enabling change, um, in my mind. And so it’s actually, once you have some part of a system it’s easier to make bigger changes than having no system at all. An example that of that is a couple of years ago. Probably like three years ago now I worked on introducing a color system to get ahead of where we didn’t really have one. We had a very small handful of, of variables, like sass variables. And we had over 2000 like hex values that were just like not connected in any way. And so now that we have some system that yeah, enables us to do, we have a system of variables we have now, it’s more like we have a handful of, of. Of values that are outside the system. The majority of the UI is using the system variables that makes it easier for us to do more work on that system. So now that we’re introducing something like working on something like doc mode and building another layer of abstraction that is far easier than starting with nothing at all. But yes, it does mean like making updates to the API of the system. And so that does mean like breaking changes at times. And so what we have to do is we have to do like an amount of a fair amount of record back to work, but we also prioritize that in, in a few different ways. With like visual refreshes, for example, we’ll prioritize that around all like most visible core components. Like when I call primary buttons, most highly visited, like iconic sort of parts of our UI. So like on most highly visited pages, like the repo. Iconic parts of the UI, like the contributions graph, the lots of sort of, we lived in looking at things like, what did the snippets of get hub UI that people share on Twitter, for example, to help us like, get a sense of that priority. Because if we try to be like, Oh, we have to do it. We think all in one go, we’re like, we’re not going to enable ourselves to sort of like iterate and incrementally progress for you, this stuff. And like learn from it along the way. So another way that we tackle sort of these risk factors is like, you know, we do look for shortcuts while I was training. You know, I think systems teams generally trying to. Fill it for the long term, they’re looking at things very holistically, whereas feature teams that are working towards sort of different goals and timelines, we still also look for shortcuts like shortcuts and like interim fixes that we can do. And the metaphor I can think of is things like codeine. Um, so we just did like a update to our icon system. And while we actually. Well, that’s a really reflected an updated, like the, the, the instances of, of those icons in get her.com. There were a ton of cases where we made our best guess and just match the older iPhone to any icon. And because we had done that in combination with spending a lot more time, like fine combing fine-tooth combing through like our client pots of UI big Mo like highly visited pages. Yeah. Then we have more confidence that the, we already have the UI changes are going to look good. They’re going to, it’s gonna. That it’s going to be not very highly visible bugs and they, because we do things like trying to iterate and ship this internally and gradually ship this to customers, we’ll pick up the sort of dark corners that we’ve missed and they sort of smaller UI bugs along the way. So even with like visual refreshes and we still are able to like prioritize and break things up incrementally.
Michaela: [00:32:13] Yeah. That’s really cool. So. You’re talking a lot about refactoring. And when I hear that, I’m thinking about tests, that’s the first thing that I was thinking about, like the safety net that you have with regression paths so that, you know, you’re not, you know, screwing something up. And, and another thing that comes to my mind is code reviews. Right. Or having other people look at the changes that you’re introducing, if they’re not breaking anything. So how is that for designers? Do you have a software engineering? Practices like that. I heard you saying something about code reviews, but it seems like more, you’re looking at the code reviews of others, but how does that work for, for designers or if you’re changing your design systems, do you have like contributes or do you have like tests that you’re, that you ran regression tests, automated tests and yeah. How would that look like for a refactoring of all of those iconic pages? I imagined this is like high risk thing. So you don’t want to break anything here.
Diana: [00:33:10] Yeah. I mean, honestly, like I think we’re finding some gaps in that we want to build more testing infrastructure so that we have a higher level of confidence in that some of that testing can be a bit less manual. I listened to your. I think you had, was it ? Yeah, so I, that was really, that was really interesting to listen to you. And, um, you know, yes, we, we were trying to find the places that we do have tests. Like we have, it’s easier for us. It’s not too difficult for us to have sort of like unit testing and stuff on our actual component libraries. But when you get to like, get hub.com. Being able to have tests to like seeing that sort of visual regression is, is something that I think there is room for improvement for, I get hub. They still like a fair amount of sort of. Manual element to that work. And that’s, you know, definitely a concern when we’re doing sort of invisible changes or changes that should be invisible like that. The user shouldn’t notice it’s really our sort of the infrastructure behind the scenes. And so, yeah, we do do a lot of code review that we do have lenses. Does that help us like spot, I guess, deviations from the system that like help us go. Okay. There’s a potential for rescue. This could bubble down and effect have unintentional, like side effects. But we do, we do have like, like a bunch of, like, we have a lot of like CIA checks and stuff that help us like catch things, help us catch things like accessibility issues. We have vessels. So you get maps into sort of the local development environment too. So you can, you can run fences and. Run some accessibility, tooling and wiring liquid development to try and catch some of those bugs. But yeah, we also have like in person or async like design reviews too. And so just like, like odd team mates to talk about code as well. And when we can, we like to sit in when it makes sense to, I really love this kind of approach of like, Really driven, um, development. So like design needs the API before you start. So they’re like really building it. So, cause that’s sort of like the interface that people are going to use. And I feel like that’s a good way to sort of walk people through the experience of this sort of. UI component changes. And in design, what we want people to do is also walk through what’s the customer experience going to be of this change. And so I think, you know, designers have to be, you know, it’s helpful if they get sort of storytellers really is like helping the audience that they’re talking to you, which might be like other PMs leadership or their engineering managers, and get a sense of the experience that a customer might have. So, yeah, we have like regular design reviews code code review, like, like actually three of the products, but also we do meet in person to talk through stuff as well.
Michaela: [00:36:11] Very interesting. Very interesting. I love to hear you explain that to me. What about feature flags? I think probably with the user base of GitHub, it’s, it’s really nice and very practical to use feature flags and see, you know, how. How people behave. You know, you, you talked also a little bit about metrics. Like if you’re rolling out a feature and you see, like, people are opting out one of the metrics, so are you using features, flags a lot? And what are some of the other metrics that you are, for example, getting there probably depends on the under feature, but what are some of those common metrics that really are insightful to help you understand whether or not there are some problems with, with the design, with the, with the feature.
Diana: [00:36:53] Yeah, so we, yes, we definitely use feature flags a lot. Yeah. Like that’s most of them development and.com is done behind a feature flag because that gives us a level of safety because it enables us to roll, roll that back and be very intentional about who we’re adding into that feature folks. So some, sometimes PE teams are just using that with their teams as a way to sort of work behind that feature together. Um, and then sometimes we also use the feature preview function that I mentioned earlier, which actually gives a way to, like, for your customers to like opt themselves into that same feature flag, basically.
Michaela: [00:37:33] What are some of the, you know, what are some of the probes that you’re actually placing throughout your work too, to get some feedback?
Diana: [00:37:42] Yeah. So we have a, we use feature flags more for, yeah. Rolling out like a feature rather than something like an experiment. So we do have a slightly different approach, even though I might be using some of that same tooling, if we want to do an AB test. And so we have a data scientist, I get her that week. If we want to like, compare like a design to another design. So. If it’s in a conversion flow or something really important, like for example, we have a sponsor’s feature. We wouldn’t want to negatively impact sponsorships. We wouldn’t want, we wouldn’t want to reduce that metric. So individual like feature teams and, and, and then we put, you know, we do have like a set of core metrics that we. Um, watch very carefully. And so if I’m walking on it, my team is working on something that you like might bump into one of those features, then I’m likely to then be having conversations with that PM and discussing with them, like, what are the metrics that we need to measure here or watch for? Do we need to, is it going to be talking to data science? Do we want to run an AB test before we do anything? Or is it just something that we want to. Track and make sure we’re capturing metrics for, as we slowly roll this out or in case we do need to roll it back. So yeah, it depends on the feature, but yeah, we, it would be a collaboration with, with products and with data science, depending on what type of thing we would, we would changing basically.
Michaela: [00:39:24] Okay. Yeah. So maybe I want to switch a little bit the gears. Talk about a different topic, which I’m really interested in. Um, and that would be your experience in a, in a very remote first organization. Right? So I have worked as an ICU, individual contributor in remotely workplaces, but I’ve never worked there as a manager. And I, I somehow feel, or imagine it quite tricky to be a manager and people manager in a remote set up. How do you experience that and how do you. How do you get the opinion of everybody and how do you connect with people? What’s what’s the new secret?
Diana: [00:40:06] I said, if I have a secret, I think it’s a thing that is challenging. I think GitHub is fortunate to have been like a remote distributed company from, from the beginning. And so. I F you know, for me, it has been a pretty great experienced, like this was my first like full time remote role. Yeah. When I worked at, at, we did have team members that were remote. Yeah. The core design team were mostly in one place. I wasn’t the roommate person. So this was like really my first experience of that. And it, it was, it definitely took a bit of getting used to, and I’m, I’m not sure, sure. That there are things that are necessarily like harder being a people manage it, being remote than there are just like being a manager is just hard and being remote. It’s just hard. And like the sort of. The crossover of those two worlds can maybe, yeah. Exacerbate some things. I, you know, I think the hard thing is always like building connections for me. One of the practical challenges is being mindful of that different time zones. And, and trying to, if I do want FaceTime, you know, there’s sometimes that it’s just not. Really possible to get every it’s not really possible right now for me to get like every single member of my team, unless someone is great working in very undesirable and hours. So what we’re trying, you know, what we try to do? I have. They smallest sort of meetings and connections that have the async communication and Slack. And we also do racing design reviews. So we use, like, we have discussions that we use fig, which is, is pretty helpful for us and works well for us because it means that. If people are in the same time zone, they can jump in, in the same tool and be in connect because you can jump on the web. Right. And so you can like be in the same tool together, which is fantastic for designers because they can like move things around together. But on the async side, they can email each other comments and chat about stuff later. So, yeah, it’s just, it’s really just being mindful of different. Ways of working, keeping, making sure that we keep working really hard at the, the async communication, not relying on, on, on, in person meetings all the time and being too biased T to like one particular time. But yeah, it’s definitely, definitely challenging. I can’t help, but point out. That it’s been even more challenging and then than usual. And so sometimes it’s just been helpful to have casual hanging out time, like social time for people. We’ve, we’ve either just like jumping on a call and just like hanging out or having a silly game to play. You know, we did like a demo day, a couple of weeks ago, and it was just super fun. Like there was people that were in Europe that stayed up. Like we started it in Europe time to get through the, and to go all the way through like the U S time so that people could do. You know, some stuff in person, but watch definitely knew, but didn’t want to, but some people were so excited that they stayed up till like 1:00 AM, so they could like be through the whole thing. It was like a conference. And that was just great. Cause it was like a way to like, have everyone together really celebrate the work that we’re doing. So I think finding the opportunities for like a bit more human connection, it has been a bit more important because we don’t, we’re not able to sort of meet in person at the moment.
Michaela: [00:43:48] Yeah. I actually see a lot of the things that you stress here for remote work is something that’s a very good mindset to have also for your design systems, right. For the inclusive design. And so you have, I’m thinking about time zone, thinking about other people’s experiences that maybe they have like dinner now, or they have kids at home and it’s lunchtime and things like that. So, yeah, I think this is probably something that makes you very strong also on the, on that, not that part. Well, yeah, thank you so much for, for being on my podcast, being on my show. It was really fantastic to talk with you. I could go on and on, but you have to, we have to stop at one point. So thank you so much and yeah. Have a good day. Enjoy your day and hope to talk to you soon again.
Diana: [00:44:34] Yeah. Thank you so much. So I really enjoyed it and yeah, I could keep going. So thank you very much for having me on the show.
Michaela: [00:44:43] Yeah. Wonderful. Okay. Bye bye.
Diana: [00:44:45] Bye. Bye.
Michaela: [00:44:47] I hope you enjoyed another episode after sup engineering unlocked podcast. Don’t forget to subscribe and I talk to you again in two.
Diana: [00:44:57] Bye.