November 21, 2020 • 29 mins
These days, there is much hype around data science and machine learning, but what's working in it actually like? Samual MacDonald, a machine learning researcher at Max Kelsen (a Brisbane-based AI/ML consultancy) gives insight into what a machine learning job actually looks like, and how he got there.
Liking The Router so far, why not subscribe in your favourite podcast app? Check out https://router.uqcs.org/ for details.
Intro/Outro Music: Awesome Call by Kevin MacLeod
Link: https://incompetech.filmmusic.io/song/3399-awesome-call
License: http://creativecommons.org/licenses/by/4.0/
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Matt (00:00):
Welcome to The Router, the official podcast of the UQ
Computing Society, where weexplore the human side of tech.
I'm your host, Matt, and todayI'll be having a chat with
Samual McDonald from Max Kelsen,here to talk about working in
machine learning and datascience in Brisbane.
Firstly, could you introduceyourself and your background?
Samual (00:32):
Okay.
Um, my name is, uh, SamualMcDonald.
Um, I'm a machine learningresearcher, I suppose.
Um, that title seems to alwaysbe changing, I think because of
the sort of nature of the work.
Um, and, um, my background in, Isuppose the reverse order is,
(00:54):
um, uh, being a, um, aresearcher at Max Kelsen and
then I'm a master of datascience student, um, at UQ.
And then before that I was, um,a, uh, uh, an environmental and
engineering geophysicist beforethat as a geotechnical engineer,
(01:15):
before that I was a uni bum.
And, um, before that I was ahouse painter.
And so that's my background, Isuppose, back to eighteen.
Matt (01:26):
And, and I guess, um, so you said you spent, uh, I guess
one or two years during yourmaster of data science.
Samual (01:33):
Yeah, I actually, um, I actually took a, uh, three years
to do that because, um, by thetime I'd started, I was, uh, I
was in, I think I was about to26.
Um, and I, I, I think that'smature age student and, um, I
had a few life experiences thatmade me realize, I suppose, um,
(01:56):
what employers are looking for.
Um, I think if they look at yourCV, they'll look at grades, of
course, there's sort of like aneasy filter to shortlist, but
the final decision doesn'tdepend on grades.
I think, um, they also look athow long you take to complete
your degree.
Um, they look at yourexperiences, um, and I think
(02:17):
they use, uh, your experiences,thirdly, the experiences as, as
the most important final factorthat, uh, ends up deciding
whether or not you get the work.
So, um, with that in mind, like,um, thinking about uni is, um,
optimizing three things, grades,speed and experiences on top of,
(02:41):
um, grades and, um, how fast youcomplete it.
I decided to take it over threeyears, um, that would, um, allow
myself more time to do otherwork and, um, get, um, um,
because I figured if I have athree years, um, I I'd, uh, uh,
I, um, all this, um, spare timeto do all this extra work, I'd
(03:04):
come out at the end with a lotmore experience than the people
I'm competing against.
And so, um, yeah, I don't knowif I'm talking about a different
answer, a different question.
Matt (03:18):
No, that's good.
Um, so I guess your universityexperience, you mentioned you
had quite a bit of work andstuff during your, um, your
university time?
What was it like doing classesafter working for so long?
Samual (03:35):
Uh, it was at that point, I'd worked for three
years in a very, um, uh, uh, uh,competitive, um, product then.
And so I was used to big hours,so my undergrad was tough, you
know, doing an engineering.
My first degree was, um, uh, incivil engineering and, um, that
was, I found very tough comparedto, um, high school, high
(03:58):
school.
I didn't remember doing anythingreally just floated on through
and got, you know, a grade goodenough to get into an
engineering degree.
It was very much focused onthings other than work& career
at that point in my life.
Um, and, um, it was only throughmy first degree that I started
to get career focused.
Um, and so, yeah.
(04:20):
Um, I think I found that thatwas really tough, but then when
I went into the real world andstarted working, that was very,
very challenging.
Um, and I think the, like,you'll end up doing, you know,
the odd week, like you're doinglike up to 70 or maybe even 80
hours, um, uh, the odd shiftwhere, um, I, you know, you
(04:41):
would have to do 20 hour shiftsor something.
And, um, that, that was veryintense.
It wasn't always like that.
It wasn't like, um, Apple backin the day or whatever, where
they consistently do 75 hours.
But, um, I think there's a lotof companies that do still do
that.
Um, and, uh, and I think, um,three years of experience of
working really hard, like thatmade doing a master's quite
(05:02):
easy.
I remember the, um, masters ofdata science, even though I was
learning a lot of new thingsabout statistics and computer
science and quite challenging.
Um, uh, I remember that being abreeze, uh, compared to, um, the
industry.
And so, um, uh, and so actuallyit was, it felt quite relaxing,
(05:23):
um, working alone and, andstudying a lot in comparison.
And so, yeah, it was justcontrast relative.
Matt (05:31):
Yeah.
Yeah.
Fair enough.
I imagine even as a uni student,I've never done anything close
to 75, 80 hour weeks before, soI guess.
Samual (05:42):
But that's different.
You wouldn't be able to causeit's a different thing as a uni
student, you're learning andlearning such a different mode
for the brain than it is.
Um, um, doing work that's notrequiring, uh, learning as much
learning anyway, the first yearof work you'll find really odd.
I think the first year of uni isreally hard because of all the
(06:04):
changes, but, um, uh, like Ithink the first year, uh, any
degree is the hardest for theperson coming out of high
school.
Uh, on average, maybe.
Matt (06:15):
So, um, when you started working at, um, Max Kelsen, was
there anything that youruniversity experience, uh,
didn't really prepare you for orthat you weren't expecting?
Samual (06:30):
Um, I think, um, yeah, I think, uh, um, and, uh, the
Master of data science, I was inthe first cohort for that.
And I think the, the, um, thelabel data scientist, I think
even still now isn't really, um,uh, consistently defined.
(06:53):
Um, and so I think, um, that,that the master of data science
did really, really well inpreparing myself for a lot of
different things.
And it had the flexibility forme to hone in on the particular
areas that I was interested in.
For example, I chose to focusmore on statistics than, um, uh,
database type subjects, but Idid do it here, database type
(07:15):
subjects.
I think I, I did, uh, um, uh,the, uh, data mining and a
couple of subjects that, um,specifically focused on database
principles.
And then, um, there was a bigdata type analytics subject, and
, um, I felt like those subjects, um, uh, didn't prepare
students well.
(07:37):
And, uh, what I think isinvolved in, uh, and, and
working with big data, um, I,that's not something I'm
particularly focused on at work.
There are people that Max Kelsenand that, uh, do focus on data
engineering and, and, MLOps, um,which is like taking everything
(07:58):
to production.
But I think that the UQ, um, uh,data science program, and I
think the same can be said forcomputer science and, and, and
software engineering morebroadly, having looked to the
old subjects in my own opinion,from what I saw, they weren't up
to date on what's being donenow, but I, I do know that UQ
(08:21):
are working with, um, uh,organizations such as, um,
Amazon web services, AWS, and,um, uh, I think, um, uh, the
sort of, um, helping them catchup.
Matt (08:37):
Hmm.
I see.
And I guess, uh, do you know anygood resources to fill the gaps
for the things that you didn'tpick up from university?
Or was it sort of just like anyarea of work you didn't really
need to?
Samual (08:52):
Uh, I think the only way is to learn on the job.
Um, I think you can try andprepare yourself with
terminology by getting a fewcertifications and stuff.
I think Amazon is great forthat.
Um, and I think Google is greatfor that also.
Um, so, uh, those certifications, um, uh, so, um, that's
(09:12):
something that everyone isgetting into put on their CV to
make themselves more attractive.
So certs, I think, uh, in ofthemselves, um, probably on
enough, um, to be a reliabledata engineer or anything, um,
uh, but, um, I think they willhelp and, um, being aware of, um
, what are the common, uh,modern methods and, um, uh, and
(09:38):
what is all the terminology.
And, um, then I think it'sreally just about learning on
the job and, um, the same, theother big thing I think that's
missing from, uh, any degree,um, is just, uh, real world
experiences.
Like how do you interact withpeople?
There's a big difference betweenbetween doing a group project
(10:01):
and actually working, um, in acompany where there's the
complex politics and clientdemands that are on the other
changing.
Um, and then, um, uh, you know,how do you interact with your
supervisor?
Uh, how do you know when to aska question and when not to ask a
question for that, get them outto the common thing I think
(10:22):
people fail at.
I think it's important to, ifyou have a problem, come up with
a list of solutions that youthink you and, and, and come up
with five different ways ofanswering the problem, even if
they're all wrong and then go toyour manager or supervisor with
that.
But that rather than just a, aquestion, um, that helps them,
(10:44):
uh, prepare their mind to giveyou a better answer, but it also
just shows that you're justdoing your absolute best.
Um, yep.
Matt (10:52):
I see.
Um, so I guess as, as, as youstarted, like working in machine
learning research or that sortof area, um, a lot of students
choose to focus on, um, one orboth of computer science or
statistics in theirundergraduate degree to try to
(11:13):
like prepare themselves for thatsort of work.
Uh, what do you believe is moreimportant between, um, between
those two?
Samual (11:22):
Between statistics and computer science?
Yeah, I think, um, computerscience sort of subsumes
statistics, that statistics isvery much a part of it.
And, um, and I think I'm verymuch like people always say
like, and it just depends onwhat you define as machine
(11:44):
learning and, um, statistics and, and, and computer science.
And this is, and it also dependson where you're at and where you
want to go.
Um, for me, I, um, uh, and my,it was, I got very interested in
the statistics because of mygeophysics background sort of,
um, gave me interest there.
And I think civil engineering ismore of a statistics focused,
(12:06):
um, engineering branch conveyedto something like mechanical
engineering where you can relyon first principles a bit more
directly.
Um, and, um, uh, so I think, um,I got very interested in
statistics and now I found thedeep learning book by Goodfellow
and Yoshua Bengio.
And, um, I read through thatbecause I got absolutely
(12:26):
obsessed to the concept of deeplearning other, I found it very
interesting.
And so that, I think put me in aposition that had made more
attracted to the, the more, um,mathematical, um, perspectives
compared to the more, um, youknow, programmatic, um,
perspectives.
Um, I think if you, um, sothat's like, um, my personal,
(12:49):
like, um, set off of, um, Isuppose, um, uh, I've just
re-labeled as between a moremathematical approach and a more
programmatic approach comparedto a more statistical approach
versus computer science.
So that's like the terminologyI'm sticking with to answer this
question.
It's a hard question to answer.
And, um, uh, and I've alsoframed, uh, where, where, uh, I
(13:13):
was positioned when I made thedecision of which I, um, pursue,
um, and, um, uh, then, uh, whichtwo, uh, the only thing
remaining is like, which toactually pursue.
And I think ultimately everyonewill have to focus on base.
Um, and I think, um, and, and itdepends on what you want to go.
(13:34):
For me, I wanted to go into moreresearch.
I wanted to learn more about,um, how it all works and, um,
what are the possibilities?
What are the limitations, um,because if you understand the,
the possibilities andlimitations, you can reason
about, uh, the safety and thefairness and the reliability of
it, and, and talk more about,and, and that will prepare
(13:55):
people for, um, realize, um, the, uh, follow one technology.
Um, but I think, um, and I thinkeveryone should try and be aware
of the limitations of it.
And I think if you're going tobe aware of the limitations of
it, you have to be aware of themathematical perspectives.
And so I think everyone has todo the basic statistics and
everyone has to, um, uh, uh, tryand understand, uh, um, as much
(14:18):
as they reasonably can.
Um, uh, so then, um, uh, uh, andit also depends on your
personality as well.
Um, uh, so then the programmaticperspective, um, you would, I
suppose you'd want to, um, this,you, that if you have more about
saying something, that's, that'sgetting some sort of an output,
(14:39):
um, if you, if you want toactually be at the edge of, uh,
any industry, if you want to doconsulting or you very much want
to get very good at being faster, just programming something,
um, knowing how to, um, uh, workwith auto ML, knowing how to do
all the pre-processing andknowing how to, um, uh, just,
um, quickly benchmark anditerate over all the different
(15:01):
models and, and learn about the,um, churning of, uh, the hyper
parameters on a very black boxperspective.
Um, and I think he can go very,very far by knowing very little
about the mathematics.
Um, and then, uh, but if he, ifyou take that road, you'll,
you'll find yourself more in theoperations side of, of things,
(15:26):
and you'll find yourself verymuch focusing on data
engineering problems and, um,uh, and, and putting things into
production.
So it depends on what yourstrengths are and what you're
interested in.
Matt (15:38):
I see.
So I guess, yeah, it depends.
Samual (15:42):
So complicated, I guess, and I'm terrible at giving
simple answers to complicatedquestions.
Matt (15:48):
Um, it is a complicated topic, and I guess it seems like
the best thing to do is just,you know, get comfortable with
both sorts of approaches.
Samual (15:59):
Yeah, I suppose it is the answer is a cliche and
that's just, you've got to doboth.
You have to do, yeah.
If you just do the mathematics,you're not going to get far at
all.
If you just do the program, um,programmatics, uh, I don't know
if that's a word, if you just dothe programming, you're not
going to get very far at all.
Um, try and do both, and then,um, decide on what you want to
(16:20):
do more of by just by how youenjoy it, what you just do, what
you want to do and what youthink you like, and don't,
don't, don't, I guess you don'thave to make it complicated.
That's what I did, like when Idecided to learn more about the
mathematical perspectives, um,uh, I just enjoy that more.
And so I did it, it was thatsimple.
Matt (16:43):
Fair enough.
I guess I sort of want to pivotmore to like, uh, your work,
work life now.
Um, one thing I'm curious about,um, what sort of like a typical
day at Max Kelsen, what's thatsort of like?
Samual (16:59):
Um, uh, and I've, I think I've been at Max Kelsen
and almost two years now, and itfeels like every month the
company has changed so much.
It's, it's, um, there's a wholebunch of different new people
and, um, the way, um, we work isvery different.
(17:21):
Um, uh, so I guess there is notypical day, like it's always
changing.
That's the amazing thing aboutdata science and machine
learning, uh, that you'll alwaysbe working on different types of
problems, probably working withdifferent personalities, very
different personalities.
That's, um, that's something Ireally liked that there's a lot
of different sort of, um, uh,characters and, um, but it's,
(17:46):
it's, um, to me, it's, it's, I,I work mostly in research and so
, um, uh, it can be with, um,looking into, um, say, um,
genomic information and, um, um,trying to answer questions about
cancer, um, or it could be, um,uh, doing, um, blue sky research
(18:09):
and today's and deep learningand learning how to ascertain
uncertainty.
That's like my key interests.
And then, um, sometimes it couldbe just like popping in and
talking to, um, some othercompanies about how we can work
together and, um, how we canprovide services to assist them,
um, in the, um, consulting sideof things.
(18:30):
Um, so there's two, two maingroups of Max Kelsen.
There's the consulting.
And then there's the research.
I work in the research, um, the,in the consulting, it's a very
different game.
It's very time focused, very,um, intense, um, uh, uh, fast
work.
And, um, and that's more about,um, the, the problems can just
(18:52):
be so diverse.
It can be about, um, orderingpizzas, or it could be, um,
helping, um, uh, uh, in housecare, um, um, whether it's, um,
uh, visualizing, uh, um, various, uh, diseases or, um, whether
it is, um, estimating how long apatient's going to be in
intensive care for, or, um,whether it is, um, does a
(19:15):
surgeon have all the equipment,um, uh, available to do their
work or, um, and, um, and everysingle there's just like so many
different, you know, insurance,and then, um, uh, there's,
there's, uh, so many differenttypes of problems and just, all
of them are just very, verydifferent.
Um, but usually it's just likeworking, um, working with a team
(19:38):
and, um, and just every, everyday feels different.
I dunno.
I can't say what a typical dayis.
Yeah.
Matt (19:52):
What's your, can you go a bit more into detail about like
your area of research?
Samual (19:59):
Yep.
Um, uh, so, um, are you moreinterested in, uh, uh, in the
genomics or in the, um, Bayesiandeep learning?
Matt (20:08):
Um, I guess both they're both big, like, areas that
you're interested in, right.
Samual (20:13):
The two groups I work with most are so in the research
team, there's like, uh, we allsort of, um, uh, uh, work, uh,
each every area a little bitwhile we can, um, to keep things
flexible and moving.
Um, that's three main groups inthe research team.
(20:35):
Uh, there's the, um, uh, there'sthe, uh, interpretable AI group.
That's where I mostly work.
And then there's the genomicsgroup.
Uh, uh, I, I work there a lottoo, and then there's the
quantum group.
Um, and, um, uh, so I can onlyreally speak for the genomics
(20:55):
and the interpretable AI, theinterpretable AI is more about,
um, uh, okay, we've got this,uh, all these new algorithms
that are becoming very powerfulbecause of the advent of GPUs
and, uh, and, um, and, and, uh,computing around big data.
So now we've put the possibilityof, um, uh, learning from large
(21:16):
amounts of data.
So deep learning is becoming athing.
And, um, but the problem is, isthat it's got this amazing
predictive power.
Um, but, um, uh, it's not, wecan't really rely on it.
We don't know when it's going towork when it's not going to
work.
And we don't know if withouttrue labels, we don't know when
it is working when it isn'tworking reliably.
(21:38):
And so, uh, that's whatmotivates the use of uncertainty
, uh, or the, the modeling ofuncertainty.
And then, uh, so ininterpretable AI, we focused on,
um, modeling, um, uncertaintywith focus on, um, we focus on,
uh, explaining, um, uh, whichfeatures or which input, um, uh,
(22:02):
inputs are, um, uh, uh,contributing the most to the
decision of in your own network.
Um, and then also, um, we'restarting to scratch the surface
of, uh, uh, causality in thatgroup too.
And then that group sort ofhelps.
Um, um, so there's, uh, otherareas of Max Kelsen including
(22:25):
genomics to, um, use a lot ofmethods from interpretable AI
also, um, and, um, in, in, inthe genomics group, um, we, we,
uh, work to solve problems about, um, immunotherapy outcome
prediction.
So if you're given, um, uh, uh,um, some genomic information,
(22:48):
uh, can we predict whether apatient will respond positively
to some treatment and then, um,uh, then there's, um, um,
problems about cancer of unknownprimary, um, uh, which I think,
um, is, um, uh, it's alreadydifficult to treat if you have a
cancer on an unknown primary,and I think can attribute it to
(23:11):
about 5% of people who die withcancer, I think, um, and then,
um, in the genomics group, um,there's a, there's a bunch of
other things that get that goaway from, um, uh, cancer,
genomics, um, uh, we'reinterested in agriculture, we're
interested in the great barrierreef and coral, um, and, um, uh,
and I think there's Maciej theresearch lead has a big
(23:36):
background in psychiatricgenomics.
He's sort of our genomics guru.
Um, and then, um, so that's,that's sort of, it's that space
there that I've just describedwhere I spend about 90% of my
time.
Matt (23:51):
I see.
Okay.
Um, I guess this is a bit moreof a, like a, a broader
question.
Um, so these sorts of machinelearning solutions, um, I guess
if you read like the news orlike if you just Google machine
learning and things like that,you'll find a lot of things
about how it's the solution toeverything like it can solve
(24:13):
every single problem.
And, um, are there any problemsthat those sorts of approaches
don't really, um, work foranother sorts of problems that
are like it, that machinelearning approaches are better
suited to solve than others?
Samual (24:30):
Yeah, I hope I can answer this question.
Um, the, um, I think as you'llvery well know, um, machine
learning has its limitations andthat it's, um, uh, very, uh,
good if you have a very specificwell-defined task, um, that is
non-stationary.
(24:51):
Um, uh, and so, um, if you, um,uh, and, and, and beyond that,
uh, it it's completelyunreliable and useless.
Um, whereas more traditional, um, engineering disciplines would
work on using, uh, physics, um,to simulate some sort of a
system and be able to makeinference about, uh, the future
(25:15):
or, um, various scenarios thatthe system can be, um, put
through.
And, um, and those, those sortof more physics based, um, uh,
uh, um, methods are veryreliable, um, in comparison to
machine learning.
Um, but, um, uh, uh, so yeah, Ithink if you've got a
(25:37):
well-defined task and if it'snarrow machine learning is very
good, um, but there's lots of,uh, narrow, uh, uh, tasks that
can be well-defined.
And so machine learning is, um,being found everywhere.
The other big, the biggestproblem, in my opinion of, um,
machine learning and in generalis that, um, it w it will by us,
(26:01):
um, um, the, what it isexperienced on just like
everyone does.
And, um, so for example, um, ifyou, if it's trying to predict
between, um, uh, I can't anadult and 99% of the, um, uh,
data points that it trains on isa dog.
(26:21):
It will commonly say a cat is adog.
Um, and, um, because th theclasses, uh, labels are
imbalanced and that's all finebecause we can just balance,
there's very easy ways to dealwith that problem.
But the concerning thing is, isthat often with tasks, there are
hidden variables.
For example, if you're not for,if, if, if you're looking at
(26:44):
genomic data and, um, uh, andyou're trying to predict, uh,
let's just use the example ofnext skeleton, if you're trying
to predict, um, immunotherapyoutcome, um, and, uh, you're
using, um, genomic information.
Um, if 98% of the, um, genomicinformation comes from, um,
(27:08):
Caucasian men, then it will beon, uh, on, uh, uh, uh, um,
black women or, um, uh, or any,any, um, minority, uh, group,
um, um, more generally any, um,underrepresented, uh,
(27:28):
subpopulations.
And so it can be unfair againstthose people.
And that's a really importantthing.
Um, when you think about, um,uh, where all the data's coming
from, um, and it's a reallyimportant thing for, um,
fairness, and it's a really, um,big shine, I suppose, that it's
starting to, um, be usedeverywhere.
(27:49):
And, um, a lot of, um, uh, um,groups, not just, not just, um,
racial or gender groups, um, butjust, uh, any kind of
subpopulation you can think ofthat is underrepresented, uh,
will be on fairly discriminatedagainst.
Matt (28:07):
Hmm.
And I guess, I guess that means,like, I guess the data set size
and the way that the data set iscollected is really important,
um, to have the solutions.
Samual (28:17):
Yeah.
That's one, that's one wayaround it.
There's other ways around it.
Um, uh, yeah, but I'm not goingto even start talking about
that.
Matt (28:27):
That's, that's all good.
Um, I think you've given areally good overview of
everything and I guess, likethat uni experience up to work,
um, and also, I guess a few, um,specifics about research, which
I found really cool.
Um, that's pretty much all thequestions from me.
Did you have anything you wantedto say?
Samual (28:45):
Oh, just thanks very much.
Yeah.
It was a pleasure.
It's really awesome that youdoing this podcast.
I think it's really cool.
Uh, I hope I could be helped toanyone at all.
Cool.
Matt (28:57):
All right.
Thanks so much.
Samual (28:59):
Thank you, Matthew.
Matt (29:00):
That's all we have for you today.
We hope you'll join us in twoweeks for the next episode of
the router.
And until then come join ourcommunity at slack.uqcs.org.
Welcome to The Router, the official podcast of the UQ
Computing Society, where weexplore the human side of tech.
I'm your host, Matt, and todayI'll be having a chat with
Samual McDonald from Max Kelsen,here to talk about working in
machine learning and datascience in Brisbane.
Firstly, could you introduceyourself and your background?
Samual (00:32):
Okay.
Um, my name is, uh, SamualMcDonald.
Um, I'm a machine learningresearcher, I suppose.
Um, that title seems to alwaysbe changing, I think because of
the sort of nature of the work.
Um, and, um, my background in, Isuppose the reverse order is,
(00:54):
um, uh, being a, um, aresearcher at Max Kelsen and
then I'm a master of datascience student, um, at UQ.
And then before that I was, um,a, uh, uh, an environmental and
engineering geophysicist beforethat as a geotechnical engineer,
(01:15):
before that I was a uni bum.
And, um, before that I was ahouse painter.
And so that's my background, Isuppose, back to eighteen.
Matt (01:26):
And, and I guess, um, so you said you spent, uh, I guess
one or two years during yourmaster of data science.
Samual (01:33):
Yeah, I actually, um, I actually took a, uh, three years
to do that because, um, by thetime I'd started, I was, uh, I
was in, I think I was about to26.
Um, and I, I, I think that'smature age student and, um, I
had a few life experiences thatmade me realize, I suppose, um,
(01:56):
what employers are looking for.
Um, I think if they look at yourCV, they'll look at grades, of
course, there's sort of like aneasy filter to shortlist, but
the final decision doesn'tdepend on grades.
I think, um, they also look athow long you take to complete
your degree.
Um, they look at yourexperiences, um, and I think
(02:17):
they use, uh, your experiences,thirdly, the experiences as, as
the most important final factorthat, uh, ends up deciding
whether or not you get the work.
So, um, with that in mind, like,um, thinking about uni is, um,
optimizing three things, grades,speed and experiences on top of,
(02:41):
um, grades and, um, how fast youcomplete it.
I decided to take it over threeyears, um, that would, um, allow
myself more time to do otherwork and, um, get, um, um,
because I figured if I have athree years, um, I I'd, uh, uh,
I, um, all this, um, spare timeto do all this extra work, I'd
(03:04):
come out at the end with a lotmore experience than the people
I'm competing against.
And so, um, yeah, I don't knowif I'm talking about a different
answer, a different question.
Matt (03:18):
No, that's good.
Um, so I guess your universityexperience, you mentioned you
had quite a bit of work andstuff during your, um, your
university time?
What was it like doing classesafter working for so long?
Samual (03:35):
Uh, it was at that point, I'd worked for three
years in a very, um, uh, uh, uh,competitive, um, product then.
And so I was used to big hours,so my undergrad was tough, you
know, doing an engineering.
My first degree was, um, uh, incivil engineering and, um, that
was, I found very tough comparedto, um, high school, high
(03:58):
school.
I didn't remember doing anythingreally just floated on through
and got, you know, a grade goodenough to get into an
engineering degree.
It was very much focused onthings other than work& career
at that point in my life.
Um, and, um, it was only throughmy first degree that I started
to get career focused.
Um, and so, yeah.
(04:20):
Um, I think I found that thatwas really tough, but then when
I went into the real world andstarted working, that was very,
very challenging.
Um, and I think the, like,you'll end up doing, you know,
the odd week, like you're doinglike up to 70 or maybe even 80
hours, um, uh, the odd shiftwhere, um, I, you know, you
(04:41):
would have to do 20 hour shiftsor something.
And, um, that, that was veryintense.
It wasn't always like that.
It wasn't like, um, Apple backin the day or whatever, where
they consistently do 75 hours.
But, um, I think there's a lotof companies that do still do
that.
Um, and, uh, and I think, um,three years of experience of
working really hard, like thatmade doing a master's quite
(05:02):
easy.
I remember the, um, masters ofdata science, even though I was
learning a lot of new thingsabout statistics and computer
science and quite challenging.
Um, uh, I remember that being abreeze, uh, compared to, um, the
industry.
And so, um, uh, and so actuallyit was, it felt quite relaxing,
(05:23):
um, working alone and, andstudying a lot in comparison.
And so, yeah, it was justcontrast relative.
Matt (05:31):
Yeah.
Yeah.
Fair enough.
I imagine even as a uni student,I've never done anything close
to 75, 80 hour weeks before, soI guess.
Samual (05:42):
But that's different.
You wouldn't be able to causeit's a different thing as a uni
student, you're learning andlearning such a different mode
for the brain than it is.
Um, um, doing work that's notrequiring, uh, learning as much
learning anyway, the first yearof work you'll find really odd.
I think the first year of uni isreally hard because of all the
(06:04):
changes, but, um, uh, like Ithink the first year, uh, any
degree is the hardest for theperson coming out of high
school.
Uh, on average, maybe.
Matt (06:15):
So, um, when you started working at, um, Max Kelsen, was
there anything that youruniversity experience, uh,
didn't really prepare you for orthat you weren't expecting?
Samual (06:30):
Um, I think, um, yeah, I think, uh, um, and, uh, the
Master of data science, I was inthe first cohort for that.
And I think the, the, um, thelabel data scientist, I think
even still now isn't really, um,uh, consistently defined.
(06:53):
Um, and so I think, um, that,that the master of data science
did really, really well inpreparing myself for a lot of
different things.
And it had the flexibility forme to hone in on the particular
areas that I was interested in.
For example, I chose to focusmore on statistics than, um, uh,
database type subjects, but Idid do it here, database type
(07:15):
subjects.
I think I, I did, uh, um, uh,the, uh, data mining and a
couple of subjects that, um,specifically focused on database
principles.
And then, um, there was a bigdata type analytics subject, and
, um, I felt like those subjects, um, uh, didn't prepare
students well.
(07:37):
And, uh, what I think isinvolved in, uh, and, and
working with big data, um, I,that's not something I'm
particularly focused on at work.
There are people that Max Kelsenand that, uh, do focus on data
engineering and, and, MLOps, um,which is like taking everything
(07:58):
to production.
But I think that the UQ, um, uh,data science program, and I
think the same can be said forcomputer science and, and, and
software engineering morebroadly, having looked to the
old subjects in my own opinion,from what I saw, they weren't up
to date on what's being donenow, but I, I do know that UQ
(08:21):
are working with, um, uh,organizations such as, um,
Amazon web services, AWS, and,um, uh, I think, um, uh, the
sort of, um, helping them catchup.
Matt (08:37):
Hmm.
I see.
And I guess, uh, do you know anygood resources to fill the gaps
for the things that you didn'tpick up from university?
Or was it sort of just like anyarea of work you didn't really
need to?
Samual (08:52):
Uh, I think the only way is to learn on the job.
Um, I think you can try andprepare yourself with
terminology by getting a fewcertifications and stuff.
I think Amazon is great forthat.
Um, and I think Google is greatfor that also.
Um, so, uh, those certifications, um, uh, so, um, that's
(09:12):
something that everyone isgetting into put on their CV to
make themselves more attractive.
So certs, I think, uh, in ofthemselves, um, probably on
enough, um, to be a reliabledata engineer or anything, um,
uh, but, um, I think they willhelp and, um, being aware of, um
, what are the common, uh,modern methods and, um, uh, and
(09:38):
what is all the terminology.
And, um, then I think it'sreally just about learning on
the job and, um, the same, theother big thing I think that's
missing from, uh, any degree,um, is just, uh, real world
experiences.
Like how do you interact withpeople?
There's a big difference betweenbetween doing a group project
(10:01):
and actually working, um, in acompany where there's the
complex politics and clientdemands that are on the other
changing.
Um, and then, um, uh, you know,how do you interact with your
supervisor?
Uh, how do you know when to aska question and when not to ask a
question for that, get them outto the common thing I think
(10:22):
people fail at.
I think it's important to, ifyou have a problem, come up with
a list of solutions that youthink you and, and, and come up
with five different ways ofanswering the problem, even if
they're all wrong and then go toyour manager or supervisor with
that.
But that rather than just a, aquestion, um, that helps them,
(10:44):
uh, prepare their mind to giveyou a better answer, but it also
just shows that you're justdoing your absolute best.
Um, yep.
Matt (10:52):
I see.
Um, so I guess as, as, as youstarted, like working in machine
learning research or that sortof area, um, a lot of students
choose to focus on, um, one orboth of computer science or
statistics in theirundergraduate degree to try to
(11:13):
like prepare themselves for thatsort of work.
Uh, what do you believe is moreimportant between, um, between
those two?
Samual (11:22):
Between statistics and computer science?
Yeah, I think, um, computerscience sort of subsumes
statistics, that statistics isvery much a part of it.
And, um, and I think I'm verymuch like people always say
like, and it just depends onwhat you define as machine
(11:44):
learning and, um, statistics and, and, and computer science.
And this is, and it also dependson where you're at and where you
want to go.
Um, for me, I, um, uh, and my,it was, I got very interested in
the statistics because of mygeophysics background sort of,
um, gave me interest there.
And I think civil engineering ismore of a statistics focused,
(12:06):
um, engineering branch conveyedto something like mechanical
engineering where you can relyon first principles a bit more
directly.
Um, and, um, uh, so I think, um,I got very interested in
statistics and now I found thedeep learning book by Goodfellow
and Yoshua Bengio.
And, um, I read through thatbecause I got absolutely
(12:26):
obsessed to the concept of deeplearning other, I found it very
interesting.
And so that, I think put me in aposition that had made more
attracted to the, the more, um,mathematical, um, perspectives
compared to the more, um, youknow, programmatic, um,
perspectives.
Um, I think if you, um, sothat's like, um, my personal,
(12:49):
like, um, set off of, um, Isuppose, um, uh, I've just
re-labeled as between a moremathematical approach and a more
programmatic approach comparedto a more statistical approach
versus computer science.
So that's like the terminologyI'm sticking with to answer this
question.
It's a hard question to answer.
And, um, uh, and I've alsoframed, uh, where, where, uh, I
(13:13):
was positioned when I made thedecision of which I, um, pursue,
um, and, um, uh, then, uh, whichtwo, uh, the only thing
remaining is like, which toactually pursue.
And I think ultimately everyonewill have to focus on base.
Um, and I think, um, and, and itdepends on what you want to go.
(13:34):
For me, I wanted to go into moreresearch.
I wanted to learn more about,um, how it all works and, um,
what are the possibilities?
What are the limitations, um,because if you understand the,
the possibilities andlimitations, you can reason
about, uh, the safety and thefairness and the reliability of
it, and, and talk more about,and, and that will prepare
(13:55):
people for, um, realize, um, the, uh, follow one technology.
Um, but I think, um, and I thinkeveryone should try and be aware
of the limitations of it.
And I think if you're going tobe aware of the limitations of
it, you have to be aware of themathematical perspectives.
And so I think everyone has todo the basic statistics and
everyone has to, um, uh, uh, tryand understand, uh, um, as much
(14:18):
as they reasonably can.
Um, uh, so then, um, uh, uh, andit also depends on your
personality as well.
Um, uh, so then the programmaticperspective, um, you would, I
suppose you'd want to, um, this,you, that if you have more about
saying something, that's, that'sgetting some sort of an output,
(14:39):
um, if you, if you want toactually be at the edge of, uh,
any industry, if you want to doconsulting or you very much want
to get very good at being faster, just programming something,
um, knowing how to, um, uh, workwith auto ML, knowing how to do
all the pre-processing andknowing how to, um, uh, just,
um, quickly benchmark anditerate over all the different
(15:01):
models and, and learn about the,um, churning of, uh, the hyper
parameters on a very black boxperspective.
Um, and I think he can go very,very far by knowing very little
about the mathematics.
Um, and then, uh, but if he, ifyou take that road, you'll,
you'll find yourself more in theoperations side of, of things,
(15:26):
and you'll find yourself verymuch focusing on data
engineering problems and, um,uh, and, and putting things into
production.
So it depends on what yourstrengths are and what you're
interested in.
Matt (15:38):
I see.
So I guess, yeah, it depends.
Samual (15:42):
So complicated, I guess, and I'm terrible at giving
simple answers to complicatedquestions.
Matt (15:48):
Um, it is a complicated topic, and I guess it seems like
the best thing to do is just,you know, get comfortable with
both sorts of approaches.
Samual (15:59):
Yeah, I suppose it is the answer is a cliche and
that's just, you've got to doboth.
You have to do, yeah.
If you just do the mathematics,you're not going to get far at
all.
If you just do the program, um,programmatics, uh, I don't know
if that's a word, if you just dothe programming, you're not
going to get very far at all.
Um, try and do both, and then,um, decide on what you want to
(16:20):
do more of by just by how youenjoy it, what you just do, what
you want to do and what youthink you like, and don't,
don't, don't, I guess you don'thave to make it complicated.
That's what I did, like when Idecided to learn more about the
mathematical perspectives, um,uh, I just enjoy that more.
And so I did it, it was thatsimple.
Matt (16:43):
Fair enough.
I guess I sort of want to pivotmore to like, uh, your work,
work life now.
Um, one thing I'm curious about,um, what sort of like a typical
day at Max Kelsen, what's thatsort of like?
Samual (16:59):
Um, uh, and I've, I think I've been at Max Kelsen
and almost two years now, and itfeels like every month the
company has changed so much.
It's, it's, um, there's a wholebunch of different new people
and, um, the way, um, we work isvery different.
(17:21):
Um, uh, so I guess there is notypical day, like it's always
changing.
That's the amazing thing aboutdata science and machine
learning, uh, that you'll alwaysbe working on different types of
problems, probably working withdifferent personalities, very
different personalities.
That's, um, that's something Ireally liked that there's a lot
of different sort of, um, uh,characters and, um, but it's,
(17:46):
it's, um, to me, it's, it's, I,I work mostly in research and so
, um, uh, it can be with, um,looking into, um, say, um,
genomic information and, um, um,trying to answer questions about
cancer, um, or it could be, um,uh, doing, um, blue sky research
(18:09):
and today's and deep learningand learning how to ascertain
uncertainty.
That's like my key interests.
And then, um, sometimes it couldbe just like popping in and
talking to, um, some othercompanies about how we can work
together and, um, how we canprovide services to assist them,
um, in the, um, consulting sideof things.
(18:30):
Um, so there's two, two maingroups of Max Kelsen.
There's the consulting.
And then there's the research.
I work in the research, um, the,in the consulting, it's a very
different game.
It's very time focused, very,um, intense, um, uh, uh, fast
work.
And, um, and that's more about,um, the, the problems can just
(18:52):
be so diverse.
It can be about, um, orderingpizzas, or it could be, um,
helping, um, uh, uh, in housecare, um, um, whether it's, um,
uh, visualizing, uh, um, various, uh, diseases or, um, whether
it is, um, estimating how long apatient's going to be in
intensive care for, or, um,whether it is, um, does a
(19:15):
surgeon have all the equipment,um, uh, available to do their
work or, um, and, um, and everysingle there's just like so many
different, you know, insurance,and then, um, uh, there's,
there's, uh, so many differenttypes of problems and just, all
of them are just very, verydifferent.
Um, but usually it's just likeworking, um, working with a team
(19:38):
and, um, and just every, everyday feels different.
I dunno.
I can't say what a typical dayis.
Yeah.
Matt (19:52):
What's your, can you go a bit more into detail about like
your area of research?
Samual (19:59):
Yep.
Um, uh, so, um, are you moreinterested in, uh, uh, in the
genomics or in the, um, Bayesiandeep learning?
Matt (20:08):
Um, I guess both they're both big, like, areas that
you're interested in, right.
Samual (20:13):
The two groups I work with most are so in the research
team, there's like, uh, we allsort of, um, uh, uh, work, uh,
each every area a little bitwhile we can, um, to keep things
flexible and moving.
Um, that's three main groups inthe research team.
(20:35):
Uh, there's the, um, uh, there'sthe, uh, interpretable AI group.
That's where I mostly work.
And then there's the genomicsgroup.
Uh, uh, I, I work there a lottoo, and then there's the
quantum group.
Um, and, um, uh, so I can onlyreally speak for the genomics
(20:55):
and the interpretable AI, theinterpretable AI is more about,
um, uh, okay, we've got this,uh, all these new algorithms
that are becoming very powerfulbecause of the advent of GPUs
and, uh, and, um, and, and, uh,computing around big data.
So now we've put the possibilityof, um, uh, learning from large
(21:16):
amounts of data.
So deep learning is becoming athing.
And, um, but the problem is, isthat it's got this amazing
predictive power.
Um, but, um, uh, it's not, wecan't really rely on it.
We don't know when it's going towork when it's not going to
work.
And we don't know if withouttrue labels, we don't know when
it is working when it isn'tworking reliably.
(21:38):
And so, uh, that's whatmotivates the use of uncertainty
, uh, or the, the modeling ofuncertainty.
And then, uh, so ininterpretable AI, we focused on,
um, modeling, um, uncertaintywith focus on, um, we focus on,
uh, explaining, um, uh, whichfeatures or which input, um, uh,
(22:02):
inputs are, um, uh, uh,contributing the most to the
decision of in your own network.
Um, and then also, um, we'restarting to scratch the surface
of, uh, uh, causality in thatgroup too.
And then that group sort ofhelps.
Um, um, so there's, uh, otherareas of Max Kelsen including
(22:25):
genomics to, um, use a lot ofmethods from interpretable AI
also, um, and, um, in, in, inthe genomics group, um, we, we,
uh, work to solve problems about, um, immunotherapy outcome
prediction.
So if you're given, um, uh, uh,um, some genomic information,
(22:48):
uh, can we predict whether apatient will respond positively
to some treatment and then, um,uh, then there's, um, um,
problems about cancer of unknownprimary, um, uh, which I think,
um, is, um, uh, it's alreadydifficult to treat if you have a
cancer on an unknown primary,and I think can attribute it to
(23:11):
about 5% of people who die withcancer, I think, um, and then,
um, in the genomics group, um,there's a, there's a bunch of
other things that get that goaway from, um, uh, cancer,
genomics, um, uh, we'reinterested in agriculture, we're
interested in the great barrierreef and coral, um, and, um, uh,
and I think there's Maciej theresearch lead has a big
(23:36):
background in psychiatricgenomics.
He's sort of our genomics guru.
Um, and then, um, so that's,that's sort of, it's that space
there that I've just describedwhere I spend about 90% of my
time.
Matt (23:51):
I see.
Okay.
Um, I guess this is a bit moreof a, like a, a broader
question.
Um, so these sorts of machinelearning solutions, um, I guess
if you read like the news orlike if you just Google machine
learning and things like that,you'll find a lot of things
about how it's the solution toeverything like it can solve
(24:13):
every single problem.
And, um, are there any problemsthat those sorts of approaches
don't really, um, work foranother sorts of problems that
are like it, that machinelearning approaches are better
suited to solve than others?
Samual (24:30):
Yeah, I hope I can answer this question.
Um, the, um, I think as you'llvery well know, um, machine
learning has its limitations andthat it's, um, uh, very, uh,
good if you have a very specificwell-defined task, um, that is
non-stationary.
(24:51):
Um, uh, and so, um, if you, um,uh, and, and, and beyond that,
uh, it it's completelyunreliable and useless.
Um, whereas more traditional, um, engineering disciplines would
work on using, uh, physics, um,to simulate some sort of a
system and be able to makeinference about, uh, the future
(25:15):
or, um, various scenarios thatthe system can be, um, put
through.
And, um, and those, those sortof more physics based, um, uh,
uh, um, methods are veryreliable, um, in comparison to
machine learning.
Um, but, um, uh, uh, so yeah, Ithink if you've got a
(25:37):
well-defined task and if it'snarrow machine learning is very
good, um, but there's lots of,uh, narrow, uh, uh, tasks that
can be well-defined.
And so machine learning is, um,being found everywhere.
The other big, the biggestproblem, in my opinion of, um,
machine learning and in generalis that, um, it w it will by us,
(26:01):
um, um, the, what it isexperienced on just like
everyone does.
And, um, so for example, um, ifyou, if it's trying to predict
between, um, uh, I can't anadult and 99% of the, um, uh,
data points that it trains on isa dog.
(26:21):
It will commonly say a cat is adog.
Um, and, um, because th theclasses, uh, labels are
imbalanced and that's all finebecause we can just balance,
there's very easy ways to dealwith that problem.
But the concerning thing is, isthat often with tasks, there are
hidden variables.
For example, if you're not for,if, if, if you're looking at
(26:44):
genomic data and, um, uh, andyou're trying to predict, uh,
let's just use the example ofnext skeleton, if you're trying
to predict, um, immunotherapyoutcome, um, and, uh, you're
using, um, genomic information.
Um, if 98% of the, um, genomicinformation comes from, um,
(27:08):
Caucasian men, then it will beon, uh, on, uh, uh, uh, um,
black women or, um, uh, or any,any, um, minority, uh, group,
um, um, more generally any, um,underrepresented, uh,
(27:28):
subpopulations.
And so it can be unfair againstthose people.
And that's a really importantthing.
Um, when you think about, um,uh, where all the data's coming
from, um, and it's a reallyimportant thing for, um,
fairness, and it's a really, um,big shine, I suppose, that it's
starting to, um, be usedeverywhere.
(27:49):
And, um, a lot of, um, uh, um,groups, not just, not just, um,
racial or gender groups, um, butjust, uh, any kind of
subpopulation you can think ofthat is underrepresented, uh,
will be on fairly discriminatedagainst.
Matt (28:07):
Hmm.
And I guess, I guess that means,like, I guess the data set size
and the way that the data set iscollected is really important,
um, to have the solutions.
Samual (28:17):
Yeah.
That's one, that's one wayaround it.
There's other ways around it.
Um, uh, yeah, but I'm not goingto even start talking about
that.
Matt (28:27):
That's, that's all good.
Um, I think you've given areally good overview of
everything and I guess, likethat uni experience up to work,
um, and also, I guess a few, um,specifics about research, which
I found really cool.
Um, that's pretty much all thequestions from me.
Did you have anything you wantedto say?
Samual (28:45):
Oh, just thanks very much.
Yeah.
It was a pleasure.
It's really awesome that youdoing this podcast.
I think it's really cool.
Uh, I hope I could be helped toanyone at all.
Cool.
Matt (28:57):
All right.
Thanks so much.
Samual (28:59):
Thank you, Matthew.
Matt (29:00):
That's all we have for you today.
We hope you'll join us in twoweeks for the next episode of
the router.
And until then come join ourcommunity at slack.uqcs.org.
Popular Podcasts
24/7 News: The Latest
The latest news in 4 minutes updated every hour, every day.
Therapy Gecko
An unlicensed lizard psychologist travels the universe talking to strangers about absolutely nothing. TO CALL THE GECKO: follow me on https://www.twitch.tv/lyleforever to get a notification for when I am taking calls. I am usually live Mondays, Wednesdays, and Fridays but lately a lot of other times too. I am a gecko.
The Joe Rogan Experience
The official podcast of comedian Joe Rogan.