A bit of AI- S02 - Ep 6

Hi everyone and welcome to the sixth episode of Season 2 of the Abitof AI show

And in this episode we will talk with B about her life in AI

Hi, everyone and welcome to the sixth episode of Season 2 of the A Bit of AI show

This episode we will talk with me about her life in AI. My name is Hank Bowman. I'm a white male with brown hair wearing glasses

And today I'm wearing a black shirt with the text, don't accept the defaults, acquired by Abel Wang

Hi, everyone. My name's Amy. I'm a female with long blonde hair that's curled

and I'm wearing a navy blue t-shirt. And actually, we do have to call out, Hank, we had a rule

We said no dark clothing, because we actually have a dark background as a visual

But hopefully it's not too bad if you are watching the actual visual version of this

Let us know. Give us feedback. Exactly. But yeah, it's still looking quite well

It's still looking quite well. So once again, everyone, welcome to the bit of AI show

And this show is all about the story from the people behind the AI systems

Because in our job, Amy and me, we meet so many people from around the globe

and we realize that there are so many different skilled people involved in creating an AI solution

from pre-sales and consultancy to creating the deep neural networks itself and running them in production

So in this show we invite people from all over the world that are professionals in the AI space

to have a chat and talk what they actually do during the week

This is the sixth episode of Season 2, and we couldn't be more excited to start talking to our guests and learn more about what they do

And as always, everything is available on our website, A BitofaI.Show. So I think we can get started and let's invite B to the screen

Thank you. Hi, Hank. How are you? Hi, Amy. Hey, thank you so much for joining us. We really appreciate it

Also, wonderful background. Love the environment that you're sitting in. Adds a lot of niceness to our show. Also, we all color coordinated. Look at that. There we go. What a weird

I know. I didn't get the old dark. That reminds me I should probably describe what I look like. So I'm a one

white female with long brown hair that is curled today. I'm also wearing dark clothes because I also did not get the memo of no dark clothes in the show

So I'm wearing black and gray. So yeah, nice to be here in the show with you guys

Oh, no, thank you for coming. Thank you for spending time with us

And yet for our guests, there are no rules. You get to wear whatever you want, which we really appreciate it

Thank you so much. But let's jump straight. in because people are wondering who you are. We are both very well acquainted with you because we're all on the same team here at Microsoft. However, do tell our audience who you are and what you do in the AI space

All right. So I'm an advocate, a cloud advocate, just like you too. I work on Hank's team, on the same team as Hank, which is the AIML team. And in advocacy, we typically say that there's like three pillars

that we focus on. There's one pillar where we connect with the product team. And in my case

I connect with the Azure ML product team. So that's my focus. On another pillar is we connect

with the community. So we connect with you guys and we hear what you have to say. We help you solve

problems. We try to kind of get you acquainted with our product. And on our third pillar

we produce content. And so we could be writing MS learn tutorials or other types of content that will help you learn

the Microsoft technologies that we work on. And in our case, even like just machine learning in general, not just as it relates to Microsoft

So that's what I do. Those are like the three things that I focus on

And so. Oh, wonderful. Well, that actually leads us quite nicely into the next question, because I don't know how I would describe this

And we ask every guest that comes on air, and I always feel a bit like a fraud, because I'm like, I'm not sure how I would describe my average day

But given that each day is different, how would you describe what your average day is

What does it look like in your role? Oh, my average day, I'm sure, I just think there's no average

But I can give you an example or examples of things that I do throughout the day

Every day is different. So every day we'll have a component of one of those three pillars

And also a lot of meetings, usually too many meetings. But let's say, for example, on a given day, I could be working with a product team

I could be trying some feature that isn't maybe that it's not released yet or maybe that it's just released

and still like in preview and that I can give them feedback

on the feature, I can do like create demos and like we call them like end to end scenarios

that include that feature and a bunch of other features together that kind of try to use their features in a context

similar to what the users would use them. So I may be in meetings with a team

I may be coaching, I do a lot of coding for demos and kind of trying to get their features working

I could be reading their documentation and then giving feedback on the documentation

So that would be like, that's probably of the three pillars. That's probably where I spend most of my time is in that connection with the team and building

these demos and end-to-end scenarios. The second one that I would say I spend most time would be producing content

I could be writing my blog, like blog posts about things that I learn

They could be things that are Azure machine learning specific, you know, a feature that I've been playing with

that I'm excited to tell users about, or it could be like an MS Learn module

to help people learn some technology that could, like for example, by torch or TensorFlow, like technologies that aren't necessarily

Microsoft technologies, but they are related to machine learning. And in the third pillar of connecting with the community

I could be I actually going to start co the AI show with Aisheaval which is Isigl is also on our little team Hank me and a few other people in the IML team So that will be a great way for you all

to talk to me directly in a live show. It's two hours every Friday, every other Friday

In like, it's the thing it's our 11. By R, I mean Pacific time

I'm in Seattle. 11 in the morning. So yeah, so that would be like a typical day

I would be switching between all three. I could be planning an AI show and then doing some demo code

and then meeting with a product team and giving them a ton of feedback

And then maybe I'll have some meetings related to some MS Learn

kind of module that we're working on and coordinating my team on that

So that's a typical day for me. gosh, busy, busy, busy, busy

That's always the thing, isn't it? And context switching as well. I can hear a lot of context switching going on

But to all of our guests, do check out Bee's Twitter profile

You can see it here, or if you go to our website, you should be able to click on her name and go to her Twitter profile

Because you share a lot of your blog posts. I always retweet them on your Twitter handle

They're very, very useful. There's lots of different, as you said, it's not all just Azure AI

It's actually like a lot of just sort of different machine learning algorithms, different things you've been working on and all that

So yeah, do definitely go and follow B. She's got some really, really cool stuff to share there

Wonderful. And yes, a big promotion, no problem, no problem. And a big promotion for the AI show as well

So the AI show is also on. Learn TV, so where you are right now

But if you head back Friday, early evenings, if you're here in Europe, or you can catch

all of these things on demand. If you search AI show on Channel 9, you can see all of them

So you don't always have to be live. I'm sure you could contact B via Twitter another time, but do go and check out that show

There's lots of cool stuff happening over there. Wonderful. Well, Hank, over to you because I know you like to dig more into the past. Like..

Exactly, yeah. So we all know now what you do during day, lots of meetings, lots of voting code, lots of meetings, code

But I'm very curious in how you got to the states in your career

Like where did you start? What did you study? And how did your professional career move to like being a principal advocate

Yeah, so I have a computer science degree, but back in the days, I'm a little old, if you guys may notice

And back in the days when I went to college, it wasn't really a thing to study machine learning in college

So I did take an AI class, artificial intelligence class, but that was very different from today's machine learning

It was all like rule-based systems, like knowledge-based systems and things like that, which aren't

really the focus these days. They're still around, but not as much the focus

And so I didn't really get into it until much later. I actually worked for Microsoft for a long time

a long time ago on client APIs for writing client side of applications

And so that was my focus. It was like APIs and front end and all that for a long time

And then, and then, you know, machine learning started, like people started talking about it

and I was paying attention. I just thought it was really interesting. And then there was one particular moment

that I remember just making the decision, oh man, this is just so useful, I need to study this

So this was a moment while I was traveling in China and south of China

I worked there, I worked in China for a little while at Microsoft

and I was studying Chinese and it's such a hard language. I was putting so many hours into it and studying so much

and learning to write also, so like learning all the strokes and all the characters. And I was traveling in my husband the south of China

and we were just like going from Little Village to Little Village

and just talking, just hiring a driver to take us from here to there

And I was like using my best Chinese and trying to tell them where I wanted to go

and go left and go right, and we need to eat, and can you stop here

just the basic things that I knew how to say. And at one point this taxi driver, like, talks to me

and he starts saying all these things, and I'm like, this is just too much for me. I don't understand what he said

And so I would ask him to repeat and speak slower and speak simpler, and he would just say things

I just didn't understand. And so, and it seemed important. Like he was kind of like annoyed

Like you want to tell me this thing? Look, no, no. Like, he would just say, say, I'm like, oh, so frustrated, right

So I pull out my phone and I tell him to say it to the phone

right, to repeat what he was saying. And he told, he repeated, and then the phone just translated it to me

in perfect English, right? And I was like, oh wow, well, that was way easier

than learning Chinese, right? And so I pulled the phone and I just spoke back

and I showed him the Chinese characters, which he understood a lot better than my very poor Chinese

And at that moment, I just had this kind of realization of like, I'm spending so much time and so much effort

learning this super interesting but super hard language. And there's like this technology, AI

that is solving this problem for everyone. And if I just put the same effort

that I'm putting into my own studying into AI, maybe I could bring benefits like this to everyone

not just to myself. So it felt like it scaled better. Like it was just like a much better use of my time

because it's just such a big bang for the buck to like solve these problems for everyone

And so that was like a really key moment. And I had already been thinking about going back to grad school

because it's something I had always wanted to do and didn't have an opportunity to do after college

And so this was just the perfect. And I wanted to go study applied math

and I knew that, but I didn't really know what to apply it to. And so when I started thinking about machine learning

and AI, I'm like, okay, this is like the perfect application for machine learning

So this is what I want to go back and study. So I did something that very few people do at age

which is I went back to grad school and and studied I got a master's in applied math

applied to machine learning at the University of Washington here in Seattle and it's not like a

professional master so it's not like something that you do after work or anything so I was

really the oldest one in the classroom by far everyone else was 22 but it was wonderful it was a

experience you know it kind of made me think like a young person again which is good and made lots of great friends and learned a lot in the meantime

And yeah, and that was great. So that's really when I then learned it

And I had already been learning it online and it was most people, like I did the Enduring courses

and other Coursera courses. And I had already been coding a lot

like doing little projects on my own. So when I got to grad school, was more like to take it to the next level

and also to kind of merge it with AI to understand more like the intersection with

machine or with applied math there's more like to understand what is the intersection between applied

math and machine learning and how can I explore that area a little bit better yeah and then and then

Microsoft hired me back which is wonderful well congratulations on heading back to academia

that can't have been easy but um it seems to have paid off as well which is

great because obviously, yeah, as you said, AI is incredibly popular. It's why we do the show

You know, a lot of people do want to learn and get involved and stuff like that

So, but also, just want to put out, it's so funny that you worked with APIs so much

And if we think about the less sort of, the more service style AI that we offer

so platform as a service such as Azure Cognitive Services, that's all about API

So really you've not gone that far away. You've just kind of gone to the other side of building it rather than consuming it

So that's really cool. That is so true. And I actually find that with almost everything in life that we learn

I feel like there's even if it feels like completely unrelated to our job or our main interest

there's always some learning that you take from whatever skill you acquire that helps everything else that comes after

If nothing else, it gives you a different perspective that other people have. that other people in the field may not have

So I definitely think that AI really benefits from people coming from like biology, psychology, like finance

fields that may seem that they're a little bit unrelated because people may think, oh, I studied this thing

that is not computer science or there's not math. But that's really where the benefit comes

because those people bring a different perspective and they're able to think of applications of AI to fields that I can think of because I don't have that background

And so I think it's really important to have diversity of backgrounds in this field

That is so true. So that brings us to our next question

So AI and your work all seems really, really nice. And it seems like you have a lot of fun

But there must be something really annoying that you find about your role in AI

And we want to know what that is. Oh. So I can think of actually a couple of things

So on my like everyday, like everyday life, I think the thing that annoys me the most is all the waiting

You have to wait for training and you have to wait to create an endpoint as you and I know, Hank, all the waiting that goes

on and like it there's no like immediate satisfaction like like there was a new

where I would just do something run it and I would see it visually show up right away

So yeah so sometimes I do something and well I'll run it for oh I'll run it

overnight and I'll check it again in the morning because you know it just takes

five or six hours to run or whatever. Actually I remember my my professor when

when I started my deep learning class in college you know on the first day you know there's

like, I don't know, more than 100 people in the classroom, most of the men, of course

And he says, you know, you guys should consider it to start taking up knitting because

you're going to be waiting a lot, whereas you're doing the homeworks or this class, and

there's going to be a lot of like that time where you're just going to be waiting for things

to happen. So that was kind of funny. But like on a more like deeper level, I would

say that the thing that I don't love about AI is that one thing I don't like a love about

AI is that there's there's a lot of potential for for doing things that aren't

ethic ethical or and and it's sometimes hard to see and I think it's it's really

important it's a really important area of AI that that people have been talking a

lot about but you know I have met people who have done research that seems like so

innocent at first, you know, and then it gets used for something that is totally not innocent

and that is actually detrimental to society. And you can't always see it when you're

when you're first doing it. And sometimes it's important to see it, right? It's important to see it

so you don't take it any further, but it's not always easy. So I feel like we're always

kind of walking a fine line of like, is this going to be used for good

so yeah that's such a good call I was actually chatting to someone about that literally yesterday

and I said the same thing I was like I always have that thing where I'm like what if I build

something and it turns out parts of it aren't ethical as you said it's used in a different

way than maybe I intended um I would never want to do it maliciously like it's just not in me

to want to do it maliciously I mean there are bad actors out there obviously um we know that

tech we you know that's not just AI um but yeah i i like i agree with that perspective i was yeah

having the same conversation with someone it's you know i guess working with other people

different people getting different people's perspectives beta testing with lots different users

all these different things but just reading lots and lots of reading right is always

useful around the area spot patterns we're very good at that i guess the machine learning

we appreciate a good pattern for sure wonderful well let's move on to our next round and our next round b is our quickfire questions

so in our quick fire question round we have six questions uh me and hank will alternate the different

questions and we would really like you to answer those questions um as quickly as possible

so the first thing that comes to your mind and if you can no more than a sentence or too long uh we have

in these rules before. So don't worry about it too much. But are you ready for our quickfire

questions? I am ready. Come for it. Okay. We'll start with a fairly historical one. So what was your

first computer Oh gosh Yeah It was a so my brother all the brother is a software engineer So it was a ZX spectrum which came out in 1982 and I was four at the time But I remember it like it just marked me It played a lot of games on that thing

I'm an early age. Perfect. What program language was used in the last project you worked on

Python for the last several years, it's been Python. That's the only thing I know at this point

I was going to say that we're certainly seeing a pattern there, yeah, in the answers

Wonderful. Question number three. So what's the most useful thing you've learned in AI

Most useful thing. I think to, I would say, like when using apps

to try to train them, like whatever knowledge they get about me

to try to train them well. For example, I get amazing ads on Facebook

I love Facebook for that reason, because I get like great recommendations

but it kind of takes some thinking to like to understand, okay, if I click on this thing that I don't

that is not what I want to get ads for in the future, then I don't, then Facebook is not gonna be a good place for me to be

But if, you know, if I want to say, see more, I don't know, something that interests me, then click there

So I guess they're starting to like play the system a little bit so that it works to my advantage

Yeah. Mmm, I love that. We'll have to dig more into that, I think

That's a good answer for that one. Yeah, that is kind of like never give you Spotify account to your children because then

your daily mixes are like like Porto or mixed with other video home, you're just

music, which is really weird in your car. Yeah. Good. Next question. What is your favorite event on the AI calendar

Oh, I'm so excited. I am going to Nereps this year. And it's all virtual, so I can't say that I'm going, but I'm attending. And I'm really excited about that. I cannot wait

Oh, when is that? It's in December, like second week of December

And this year it's only $175. I encourage everyone to, it's like the perfect year to go because it's so cheap because it's not in person

It's online. So I think this will help democratize access to this conference, at least this year

I don't know if they're going to go back to in person, but it's something that most people should be able to afford these days

I love that. That's such a good one. I went in, oh, I think it was like 2018. And I still, to this day, saw my favorite keynote ever by a researcher named Kate Crawford. I talked to it about everyone. If anyone has seen any of my stuff on the internet, I literally talk about this keynote so much. It's such a good conference. You'll really enjoy yourself

Oh, I can't wait. On to, I know, on to question number five. So, um, whatever

area of AI is on your list to skill up on next

I've been looking a lot into an area called like scientific machine learning, where you can use AI

to solve partial differential equations or to discover partial differential equations based on data

And so it's an area that I got a little bit into it during my master's, but not as much as I like

And so I've been looking at some things. It's been around for a while, so it's not like super recent research

but I've been looking at physics-informed neural networks, or PINs, B-I-N, and a few other related technologies

And so maybe there will be something in my blog about that at some point

We'll see. Yeah. Thank you. Cool. Looking forward to that. Last question

What was the first thing like project or scale you built in AI

So the first thing you touched when moving to the AI field

I think my first project in class, like a class project, was the eigenfaces project

It's a lot of people's first project, which helps to understand about

principal component ysis and also like image compression and things like that and it's it's yeah I think it's somewhere on my get-hah still because I think I put all my

projects from grad school there but it and it's also like something that can like it's self-contained

and small enough they can be done as a homework pretty easily so a lot of classes will use that

as a like a starting point I think that was my first at least of my first in in college I would

say oh that's a brilliant one it's really good one I think I did that

project as well certainly rings a bell um yeah yeah um so yeah they use our principal component

ysis was something i used quite a lot because i did a lot with text um to start off with so cutting down

a lot of text was quite important for that but we've come to the end of our quick fire round and

just big thank you because you gave some fascinating answers and we really really appreciate

the surprise element of the quick fire round and just big thank you because you gave some fascinating answers and we really really appreciate uh the surprise element of

the quickfire questions as well. But B, with that, we are actually out of time

Oh, my goodness. The show always goes so quickly. We really, really appreciate you coming on the show

spending 30 minutes with us, telling us all about yourself. And, yeah, hopefully people can go, follow you on social media

and see all the great stuff you're working on. But with that, yeah, thank you so much for joining us

Thank you so much. Amy and Hank for having me on the show. It was fun

Thank you. All right. We'll see you shortly. Bye. Fabulous. Hank, as always, we are at the end

This one we maybe cut a little fine. So I guess the last thing we can do is just wrap up the day

So thank you so much for joining us today. As always, it's been a pleasure to have you as our audience

If this is the first time you've caught a bit of AI, you can catch all of our on-demand episodes, including season one, on our website at A.i.com

And also, do check out the website because you can see all of our different speakers

You can get in contact with them. You can go look them up online and see the great work they are doing

And with that, we'll say thank you so much for watching. And this has been A Bit of A.I. with Henkin Amy

And we'll see you soon. See you next week