A bit of AI Ep. 3 | S2

Hello everyone and welcome to the third episode of Season 2 of the Abit of AI show

In this episode we will talk with Tanya about her life in AI

Hi everyone and welcome to the A Bit of A.Bit of A.I. Show. My name is Hank Bowman. I'm a white male with brown hair wearing glasses. And today I'm wearing a dark red hat

shirt with a Microsoft sweater which is grey. Wonderful and hi my name is Amy. I'm a female with blonde hair and

today I'm wearing a white shirt and it's got yellow hearts on it

And once again welcome everyone to the A Bit of A.I. Show and this show is all about

people that work in AI, the people behind the AI systems. Because in our

jobs Amy and me meet many people around the world and we noticed that there are so many different people involved of actually building an AI system

that is in production. So in this show we try to find out how people got to that point in their

career and what they're actually doing during the day. And in this show we will talk with

Tanya. And as always everything can be found on our website a bit of AI. Show

and as always I have a few of the logistics to share with you so after the show you're more

than welcome to join us in the A Bit of AI cafe. It's a space where we can get together after the show

and have those side conversations that we used to have when we were in person and networking and

meetups. So if you want to join us, it's straight after the show in 30 minutes time and you can go to

a.m.s. slash a bit of AI dash cafe. So yeah, hopefully we'll

We'll see you there. It's always good to chat. Cool. So let's get Tanya into the show

Hi, Taya. Hello. Hey, how are you doing? Very good, very good. It's so great to have Tanya on the show as me and Hank used to work with Tanya

which is wonderful. So it's been so good to catch up in our prep session and kind of check out what you're up to

But Tanya, they're not here to listen about who we used to work with

here to listen about you so tell us a bit about yourself well i want to introduce myself i am tanya

lard i am a female with short purple hair i am also wearing glasses and a matching light like jumper

that also has a colorful python wearing a hat it's like a bit out there um yeah no i all

since i left microsoft i was well now i am working as a co-director in quonside labs um which is

and non-profit innovation labs for open source projects. Very, very cool. And can you tell us a little bit breakdown Quonsite

Because when I went to the website and I was kind of having a good look at what

there's like two sides to the business, isn't there? Can you let us know a little bit more

what you kind of do at Quantite? That's correct. So we have Quantite as an organization

One side of it operates smoothly. as a consultancy business for data science machine learning so that means that we have clients

spying all different kinds of industries from banking to energy to research and then we have

the other side of the organization which is actually the one that i'm co-directing with ralph

gomers and that is quantite labs that's the non-profit side and our main mission is empowering and

and sustain open source projects and our community. So we do a lot of innovation, like trying to do a bit of an R&D

the kind of things to improve libraries, like build new features, build new tools

and also make sure that, well, we can keep on living all of those open source projects

that all the community, all the data science community, really day by day

I love that. I was going to say, like, yeah, really open, is so well used specifically in data science as well. And so, you know, having that kind of maintenance of these different packages and you and your team, you know, paying so much closer attention to them obviously means that they get better all the time. So that's really, really exciting. One thing we ask all of our guests because titles and what companies do and stuff like that, we all know that the day job sometimes looks a little different. So, um, what

What does an average day in your job look like, Tanya? Are you coding a lot

Are you talking to people a lot? Like, what's the balance? So I would say my day, like

it's divided in three equal parts. It's not always the same. So 30% of my time corresponds to technical work

It can be either working directly on clients projects or sending pool requests, making feature

on different open source projects, mostly like on the love side of things

or operation side of things for machine learning. Then other 30% of my time is more about strategic work

meaning the direction and leadership of labs, doing more things like funding, writing grants

so that we can continue funding the open source projects that we work with

And then the other 30% of my time is, managing basically a team of open source developers that are distributed all across the world

But of course I also do a lot of outreach like coming here speaking at different conferences I sit in different board of directors So I also try to fit that in into my non let say 10 or 20 extra pot in my job

So I have to do a lot of balancing, like all of these responsibilities

I love that. I love that. And yes, no, it's definitely a big call out

A lot of people on this show that we have on as guests are

we believe very important because they promote kind of community. and you're a huge part of Pi Ladies, right

For those, for anyone out there who wants to kind of get involved in community

Yeah, that's correct. I am currently sitting in the Pi Ladies Global Council

And, well, we're basically the top body that makes sure that all of the other chapters

because we have like so, so many chapters distributed around the world

So we provide basic infrastructure support. we help them to actually, well, if someone wants to start a new chapter in a location where

there is not a pilotess chapter, we help them with that initial starting of their

chapter. We do a lot of things around like Code of Conduct, make sure that they have all the

tools to actually run different events and initiatives across the world. That's awesome. I was going

say way back when pie ladies and our ladies i think are the two best hackathons i've ever been to like

i've just i just enjoyed them so much um yeah anyway side note go check out those communities

another community for people to go and join that we mentioned on this show you also have

and the kind of the personal side of your life you have some wonderful dogs uh do you have time to get

out and about how do you balance those time zone issues Wow. So my dog actually is a big part of what actually helps me keep my work life balance. I have to be honest, before having a dog. I could just like so easily work, super focused on something and forget about breaking or sometimes having lunch. But now my dog is very, very part of my routine. In the morning, we go out for a long walk. It can be anything from like 40 minutes to an hour

Then during lunch, we go for another walk. Sometimes I have meetings, so my significant daughter has to take the dog actually

And then in the evening, we do it the same. One of the problems is, well, not a problem, but like something that is different to other places

that I've been on is my team is highly distributed. And a lot of the people that I actually manage are in America's time zone

So now I'm starting work a bit later than I used to

I am exercising in the morning and then starting my workday, break for lunch, come back

and yeah, just crack on with stuff. Normally, afternoons are pretty busy with calls because, yeah, like team and leadership

and clients tend to be everywhere. But yeah, I think my dog really, really helps me to be focused in, like, being more

mindful about taking breaks, like exercising in the morning, not being sat down for 12 hours

sorry, 10 hours, how many hours? Or just being in the same room

Do you have, I feel like I've shared this probably loads on this show, but you know that like, especially going into winter over here in Europe

those times when it starts to get darker earlier, and then you're literally, you suddenly realize you're sitting in a dark room

with lots of screens staring at you, and you're like, whoa, I need to stand up and move here

This has been in one place for way too long. But you've heard it here first team who are watching

get a dog, it gets you into really good exercise patterns. There we go

Hank, over to you. You have some more questions about kind of how Tanya got to where she is

because she's so successful. Yeah, exactly, because I'm always really interested in how you get from where you started

You went to school somewhere. You started something to actually this point in your career

So can you tell a little bit about your journey? Right. So as many other people in this space or in tech and machine learning, I don't have like a very direct route into machine learning

So I actually started from a mechatronics engineering. Because I've always liked multidisciplinary kind of projects and things and a chance

But after I went into there and in my master's and started working, I realized that I really liked the computer science

like the programming side of things. So when it was time for me to choose what I wanted to do

a PhD in, I found a project that was, again, like, multidisciplinary

It was into material science, but it was not traditional material science kind of project

in which you are in the lab all the time and in the real samples

but it was more on pure optimization, which is like machine learning applied to bioengineering and biomaterials

So as I was doing that, I realized like, oh, like I really, really enjoy doing all of this programming stuff

I got very deep into all of the Pi data or scientific computing Python libraries

I really, really enjoyed what I was doing. I got to meet some people that were labeled as research software engineers at the University of Manchester

And I just got to talk to them a lot because of, well, I was the only one doing computation

know, it kind of work in materials. I relied a lot on their expertise and their best practices

And it was like, oh, this is something that I want to do. So when I finished my PhD, instead of going through or down the traditional academic

route of like going into lectureships, senior lecture and all of that, I just decided to go

down the research software engineering path, which has taken me to like different universities

different startups, Microsoft. of Quartet Labs and now what I do, I do machine learning for like in so many, like

I would say full-stack machine learning that goes from applied data science projects to

MLOBS, all of the operation side and like making sure that you can put things into production

how you take things from R&D into production, to actually working on down infrastructure

and tools that's both the data science ecosystems. So yeah, I think it's well override

I think I always had, as I said, I had always had this interest in, like

different projects with different scopes, different disciplines. And yeah, I finally made it

Like I get to put my fingers, many different areas and many different disciplines with machine learning

Good. Interesting. How big was the role of community in your career

part? Huge. I would say if I hadn't gotten involved in a community or like if I were

a part of the different communities I belong to, I doubt my career or my journey would look

like this way. Especially when I was doing my PhD, I used to feel very isolated because it was

so different other people or in my institute were doing So I was always a weird one that was at this intersection of computer science mathematics material science So when I found this research software engineering community and the carpentries

I was like, oh, there are other people. So I just started getting into the Python community, the R community

And I just felt like I found my tribe, like my people, because they were doing incredible

thing. We're doing a very wide variety of things with Python, again, like microcontrollers or

network staff, DevOps stuff, web, like all across. And they were just so welcoming. So I just

decided to put a lot of time in effort in being part of the community, building communities

I also have been very, very lucky because of this involvement or me being so active and

proactive. I've gotten to get like really good mentors that have helped me throughout different

periods in my career and some of them are now like real really good friends of mine. So I think I am

a lot of my career path and journey is thanks to the community. Like community has played a

massive, massive role for me to be here where I am. Great. Great to hear. Then it can't be all

good and nice and so on. So the question for me, when I always ask everyone, is what is the most annoying thing in your role in AI

I just have to choose one, right? Yeah, just... Just the one, yeah

I would say, because a lot of us folks that are working in machine learning or AI design

we're very cool and number, like the taste of their mathematics focused

But we always have to remember the data are like a lot of the data that we work with represent individuals across the world

So something that really nice is that we focus a lot on the deck or it's scaling up or the tools and infrastructure making better models

But in many, many projects, ethics is still an afterthought. ethics should be a core part of future governance processes, the same way that we are paying so much attention in choosing, which is the best framework, like, I don't know, Python or TensorFlow, or how much data we're collecting, or, like, how much we're improving the performance or accuracy of our models, which should pay as much attention to ethical compliance, like the extent implications that what we're doing will have in society and people

people and I don't think we're we have gotten to the point that ethics is the

mainstream or like the norm of every project has an ethics component I think that

bothers me a lot like that that annoys me I've seen like there of course a lot of

researchers and a lot of people out there there are being pretty good

advocates and activists in the ethics area but there is still much more that

should be done yeah oh no very very very good call out. Thanks for sharing that one. Remember when we're when we're talking to

Tanya you can join us in the cafe afterwards. I've already noted down a few things. I want to

dig into mentors with you. I want to dig into some of the amazing projects you work on

But just before we let Tanya go, we have our quick fire question round. We have six

questions for you and we would like you to answer them if possible whatever the first

first thing that comes into your mind is, so whether it's a single word or if you can

no more than a sentence long, we would really appreciate that. So, Tanya, are you ready for the

quickfire round? Okay, I think I am. I'll just have to psych me up. It's okay. I know. I'm

building it up. It's not actually that hard. It's not that big. So our first question in the

quickfire round is hopefully a nice easy one. What was your first computer? I can't even

remember that. I have no idea. It was a super, super old IBM. I am 100% sure

Can you remember how old you were? Probably 12. I wish just using my mom's computer, really

things like fragments. Cool. Then this one. What programming language was used in the last project you worked on? Python. No

There we go. That's the same answer again and again and again

But also, Tanya is an absolute Python expert. So to give context, I'm not surprised that she said Python just then

But yeah, next question. What is the most useful thing you've learned in AI

Linear regression. Yes. We love the basics. What is your favorite event on the AI calendar

Oh, not AI, but PyCli. but Pike on US. Pike on US

Very nice, very nice. Is the next one coming up or is it just been

So the next one is coming up on the 2nd of May 2020

and it covers all of the uses of Python. So that covers AI and machine learning of course

Amazing, amazing. Okay, cool. Next question. question so we're nearly at the end time you're doing really really well and

what area of AI is on your list to skill upon next so what you're looking to

learn oh um and it'll be I would do it LLP we are still a lot of text out there

to be yzed a lot of text out there Yeah. And what was the first thing, project or skill you built in AI

Oh. I think it was my PhD thesis, actually. It was a big one. So

Yeah. Can we get you to expand a little bit on that one? I'm going to break the rules a little bit on our quickfire questions

in that, was it like a package that you learned to use in AI

So like the first thing you did in machine learning? So it's the first proper project that I, that was machine learning

Of course, because it was material science, we'd never call it machine learning

but it was like a ton of optimization, a lot of optimization

So I got to you, as I said, like the whole. all Python scientific not scientific computing Python stack like things like Psychet Learn Pandas NumPy all of that amazing amazing no thank you for sharing

because it's always good i think for our we've added that question in that's a new one for season two

and we think it's quite important for those who are out there that aren't quite into machine learning

AI yet to let everyone know like is there a pattern around the first thing people learn

so yeah that's a that's a really good one non-pie all those packages

are very, very well used. And I guess quite a nice circle because I guess now you're very much

working still on those kinds of open source packages and just making them better and better

Yeah, yeah, it feels like I've been using them and now like working on them for like forever

Like it feels so long. I just can't imagine me moving to something different because I'm so used

and so familiar with this stack. And honestly, I love it. I love it. I love it

the community. I love the projects, everything. Oh, amazing. Amazing. In the last minute that we have

with you just here, Tanya, I just wanted to very briefly dig in. Do you have a process for learning

all these different packages, these different technologies? Is there a way that you go about that

I do a lot of reading of papers. I rely a lot on blogs of people that are like. I rely on blogs of people that are

like at the top of the field, I love going into papers with code

That is one of my favorite pages. And also a lot of them, the learning that I do is by doing hands-on work on specific projects as well

I think that works for me, learning by doing, mostly. Learning by doing

That's great. Well, Tanya, thank you so much. We, how quick does this show go when we're just chatting away and we have so many

questions for you. As I always say, join us in the cafe. Tanya's going to be able to join us for 30

minutes. You might have just seen me writing down some stuff. These are literally questions that I

want to ask Tanya a little bit more about. But don't let me take all of the time in the cafe

Join us. Ask your own questions and we can have a good chat after the show. But Tanya

thank you so much for joining us. It's been an absolute pleasure to talk through the amazing

work you're doing now. And hopefully people can keep in touch with you. Yeah, absolutely. Thank you for having me around

No problem. We'll see you very shortly. Okay, so in the last few minutes, Hank, we always do a bit of an announcement, don't we, as well as some of our resources

This week, we actually wanted to announce a conference that's happening on Saturday in the UK. It is an in-person conference

And the reason we want to mention it is because Tanya, our guest, is actually keynoting at it

So it's called developer, developer, or DDD East Midlands. You can go and find that at DDD East Midlands.com if you want to go and see what's on

So even if it's not something you could make in person and you're from a different country or anything like that

still go and check out what some of the subjects are, what some of the resources are, and actually the great set of community people that sit behind this conference as well

because they do so much more than just an in-person conference. So that's on Saturday, the 2nd of October from 9 a.m

And it's free, it's in-person if you're based in the UK. So yeah, good luck to Tanya

I'm sure she'll do an amazing keynote for those who are able to attend

And with that, we'll start to wrap up the show. Hank, is there anything else you're keen to share with us today

No, no, actually not today. No, I was going to say it's been, we've talked about global AI community back together last week

Go check out that. We'll just a very, very brief plug for it, globalaI.comunity

Go check out that website. Go see what it is. It's a community for everyone in AI who wants to get learning

and there's some cool stuff coming up with that as well. Okay, so we will wrap up the show today

Thank you so much for joining us. We really hope you enjoyed this show with Tanya Allard

You can catch up all episodes. Be that, we're in episode three now in this season

but you can catch all of them and season one by going to our website, AbitofaI. show

Don't forget, if I've not already said it enough in this session

Join us in the A Bit of AI Cafe, a really nice sort of after show experience where we can just talk and network and get to know each other as a community that watches the show

Go to AK.m.S. slash A Bit of AI dash cafe or go to our website and it will be in the top right corner

And with that, I want to thank you for watching. This has been A Bit of AI show with Henkin Amie