►
From YouTube: Built to grow: How the world's largest company creates a codebase that scales - GitHub Universe 2018
Description
From the the supply chain to the customer journey, Fiona Tan, SVP, Customer Technology and Artificial Intelligence at Walmart, discusses how to build on your code base to meet the demands of a growing digital economy. Learn what it really means to "deploy at scale" from the team’s experience serving over 300 million customers every week, aiding in recovery from natural disasters, supporting social causes, and more.
About GitHub Universe:
GitHub Universe is a two-day conference dedicated to the creativity and curiosity of the largest software community in the world. Sessions cover topics from team culture to open source software across industries and technologies.
For more information on GitHub Universe, check the website:
https://githubuniverse.com
B
Hello,
everyone
welcome
back
from
lunch.
My
name
is
Fiona
tan
and
I
work
at
Walmart,
Labs
I
lead
a
team
called
customer
technology
and
I
thought
I'd.
Tell
you
a
little
bit
about
how
we're
coding,
testing
and
deploying
at
scale
at
Walmart,
so
first
a
little
bit
about
Walmart
labs,
we're
based
in
actually
different
parts
of
the
country
as
well
as
in
India
and
we're
the
tech
and
innovation
organization.
B
That's
powering
the
world's
largest
company,
so
my
team
in
particular
the
customer
technology
team,
is
focused
on
customer
impacting
and
customer
facing
technology
at
Walmart,
and
we
cover
everything
across
our
ecommerce
sites.
Our
apps,
as
well
as
our
stores
I,
have
about
18,
100
or
so
developers.
Then
my
team
and
we
spread
out
across
California
in
Arkansas
Bentonville
and
Bentonville
Arkansas,
Reston
Virginia,
as
well
as
Bangalore
in
India.
B
So
what
I
thought
I'd
do
today
is
to
tell
you
a
bit
about
Walmart
scale
and
how
we're
using
tech
in
the
world's
largest
company
I'll
present
an
example
that
details
an
end-to-end
customer
journey
discuss
some
of
the
complexities
that
come
with
that
and
the
impact
that
we
have
when
we
have
quality
issues.
If
we
have
any
and
also
how
we've
adopted
a
culture
of
quality
and
develop
an
open-source
automation
platform,
that's
helping
us
with
that.
B
So
as
the
world's
largest
company
talk
to
you
a
little
bit
about
the
scale
of
Walmart,
it's
not
surprising
that
we
have
also
sort
of
the
operations
at
scale
is
somewhat
unique
to
Walmart.
So
we
have
270
million
customers
that
shop
with
us
weekly
around
the
world.
We've
got
11,000
stores
around
the
globe,
specifically
in
the
US.
We've
got
over
150
million
customers
that
shop
with
us
weekly
in
our
stores
and
that's
4,500
stores
that
we
have
across
the
country.
We
have
another
100
million
that
shop
with
us
online
and
another
trivia
point
about.
B
Ninety
percent
of
Americans
actually
live
within
a
10-mile
radius
of
a
Walmart
store
and
that's
a
you
know.
I
think
when
you're
in
San
Francisco,
you
don't
really
see
that
we're
clearly
and
the
ten
percent
of
that,
but
in
the
other
parts
of
the
country,
like
I,
said,
ninety
percent
live
actually
within
a
ten-mile
radius
of
Walmart,
so
we're
very
much
integrated
into
the
fabric
of
life,
I
think
in
America,
so
we're
also
the
world's
largest
employer.
B
We
have
over
2.2
million
associates
that
work
for
Walmart
and
just
from
a
context,
perspective
right,
I'm
from
Singapore
when
I
left
Singapore
to
come
to
the
US
to
go
to
college.
There
were
about
2.2
million
people
living
in
the
entire
island
country
right
so
I.
Think
about,
like
the
entire
country
of
Singapore,
was
working
for
Walmart,
so
just
to
kind
of
again
a
little
bit
about
the
scale
of
Walmart.
So
as
a
technologist
as
a
developer,
you
know
why
Walmart
right,
I
think
it's
it's
really
for
us.
B
It's
since,
for
me,
it's
you
know
personally,
very
inspiring
to
be
able
to
impact
the
lives
of
so
many
people,
hundreds
of
millions
of
people
around
the
globe.
It's
obviously
customers
that
depend
on
us
and
our
mission
in
terms
of
helping
them
save
money
and
live
better
is
something
that
really
resonates
with
me
and
again
on
the
flipside.
If
we
don't
do
it
right,
you
know
we
also
have
impact.
That's
you
know
somewhat
negative.
So,
like
I
said,
quality
is
is
obviously
very,
very
important
to
us,
so
the
scale
of
the
customer
technology
team.
B
So
that's
my
team
and
if
you
look
at
the
circles
and
then
sort
of
the
different
teams
that
work
there
you'll
see
that
it's
it
kind
of
spans
a
whole
spectrum
right,
so
we've
got
personalization
and
machine
learning
that
that
powers
a
lot
of
outside
it
powers
a
lot
of
the
advertising
technology
that
we
have
and
marketing
as
well,
that
sort
of
helps
drive
traffic
to
our
site.
So
there
are
teams
that
doing
that.
B
Then
we've
got
customer
experience,
teams
that
are
building
our
websites
they're
building
our
mobile
apps,
and
you
know
making
sure
that
we're
able
to
deliver
a
really
delightful
experience
for
our
customers,
and
then
we
have
teams
that
work
on
store
systems.
So
if
you
do
go
to
a
Walmart
store
the
point-of-sale
systems,
the
self
checkouts,
the
pharmacy
systems
that
are
dispensing
prescriptions
as
well
as
systems
sort
of
at
the
back
of
the
store
that
the
associates
use
everything
from
receiving
items
right
from
trucks
all
the
way
through
stocking
the
shelves.
B
So
when
you
go
to
a
Walmart
store,
you
should
hopefully
be
able
to
find
what
you're
looking
for
so
there's
a
lot
of
tech.
I
think
it
was
surprising
to
me
when
I
first
joined
at
Walmart,
how
much
tech
there
is
behind
and
that's
powering
near
the
retail
business
right.
So
all
of
that
together.
So
one
of
the
things
that
you
probably
also
you
know
realize
that
hey
look.
Actually
you
know
our
end-to-end
journey
is
spanning
everything
from
the
website
to
the
apps,
to
the
stores
and-
and
it's
not
it's
actually
really
by
design.
B
So
about
three
years
ago,
I'll
see
your
Doug
McMillon
talked
about
how
how
Walmart
was
gonna,
win
in
retail
and
really
the
way
he
talked
about
it
was
that
we
would
need
to
leverage
all
the
assets
that
we
have
right.
So
it's
the
11,000
stores
that
we
have
all
wide
it's
the
4,500
stores
that
we
have.
You
know
within
10
miles
of
most
of
the
population
of
the
US.
B
How
do
we
leverage
that,
along
with
our
digital
assets,
right
so
the
ecommerce
sites,
the
apps,
as
well
as
some
new
technology,
that
we're
working
on
right,
voice,
comments,
etc?
How
do
we
combine
all
of
that
so
that
we
can
really
provide
our
customers
with
a
very
seamless,
omni-channel
experience?
However,
she
wants
to
shop
with
us
whether
she
wants
to
go
to
the
store
whether
she
wants
to
shop
online.
B
So
really
that
provides
a
lot
of
really
interesting
opportunities
for
our
customers,
because
what
it
means
is
that
they
can
buy
online
and
pick
it
up
in
the
store.
They
can
go
to
the
store,
download
our
app
and
find
items
in
the
store
they
can
check,
inventory,
etc.
So
the
thing,
though,
about
combining
sort
of
is
physical
and
digital
aspects.
You
know
the
end-to-end
journey
does
span
a
lot
of
disparate
systems,
so
I
think
what
we'll
talk
about
a
little
bit
is
how
we
bring
our
teams
together.
B
Okay,
so
this
was
the
an
example:
this
is
the
online
grocery
order
flow,
so
something
that
we
offered
through
most
of
the
country.
You
have
Lisa
she's,
a
customer
of
ours.
She
goes
to
our
site,
she
might
use
the
website.
You
might
use
the
app
she's
adding
items
to
her
cart
and
then
she
checks
out
at
that
point,
there's
an
order
management
system
that
kind
of
takes
over
it's
sort
of
think
of
it.
B
It's
like
air
traffic
control
right,
so
basically,
when
you're
ordering
something
online
for
online
groceries,
there's
actually
a
store,
that's
associated
with
you.
That's
where
you're
gonna
be
picking
up
your
order,
so
the
order
management
system
is
actually
going
to
route
combined
and
route
orders
with
the
appropriate
store,
there's
also
a
picking
system
at
the
store
and
what
its
gonna
do.
It's
gonna
optimize
as
all
the
orders
are
coming
in
it's
trying
to
figure
out.
You
know
by
time.
When
are
these
orders
due
to
be
picked
up
and
it's
going
to
combine
orders
accordingly?
B
That's
optimized
for
the
associate
as
he
or
she
is
navigating
the
store
to
pick
up
the
orders.
So
a
pretty
interesting
stuff
right,
if
you
remember
all
your
graph
algorithms
from
school
kind
of
comes
into
effect
here.
So
all
that
needs
to
happen
before
the
customer
shows
up
right.
So
when
the
customer
is
close
to
the
store
she
checks
in
on
her
app
to.
Let
us
know
that
she's
close
by
the
associate
then
at
that
point
goes
and
picks
up
the
order
and
they
are
gonna.
B
You
know
if
all
things
are
aligned,
the
customer
pulls
up
into
the
parking
bay.
Just
as
the
associates
coming
up
with
the
order
loads
it
up
in
her
trunk
and
she's,
often
she's
a
happy
camper.
So,
as
you
can
see,
all
of
this
needs
to
be.
You
know
pretty
coordinated,
there's
a
lot
of
different
services.
They
all
need
to
talk
to
each
other.
They
are,
they
all
need
to
integrate
with
each
other
and
many
services
across
disparate
systems.
B
If
you
look
at,
you
know
the
website
and
the
app
you
know,
you've
got
Java
JavaScript
on
the
website.
Potentially,
you've
got
different
native
apps.
You've
got
Java
services
that
power
most
of
the
back
end,
and
then
you
got
legacy
systems
they're
written
in
various
different
languages
long
long
time
ago,
potentially,
but
all
of
these
systems
need
to
come
together
in
order
to
deliver
one
good
order.
So,
as
you
can
imagine
everything
everybody
would
agree
with
me.
Quality
is
very,
very
important
at
the
scale
that
Walmart
operates.
B
It's
even
more
critical
you,
you
know
we
saw
again
all
the
disparate
systems
that
need
to
work
together,
that
complexity
being
able
to
solve
these
complex
problems
at
scale.
I
think
you
quickly
realize
that
there's
no
real
there's
not
really
one
team,
one
individual
one
team
that
can
deliver
a
solution
independently
right,
we're
all
very
dependent
on
each
other.
We
need
to
really
work
in
a
very
collaborative
development
environment
and
we
need
to
find
a
way
to
double
down
on
quality,
but
not
slow.
B
The
developer
down
right
so
I
think
that's
a
big
goal
of
ours
and
you
know
we
measure
quality
in
three
key
areas.
There's
clearly
financial
impact
right.
So
if,
if
there's
a
quality
issue
and
our
site
is
down,
customers
can't
come
to
our
side,
they
can't
place
orders.
They
can't
check
out
clearly
there's
a
financial
impact
from
the
customers
from
the
from
the
company's
point
of
view,
there's
also
the
time
impact
right.
B
So
if
you
have
an
issue
in
production,
it
takes
a
lot
of
time
for
the
SRE
team,
the
dev
team,
to
triage
to
diagnose,
to
fix
clearly
a
lot
more
time
spent
doing
that.
Then,
if
you
actually
are
able
to
find
these
issues
in,
you
know
functional
testing
or
integration
testing
and
then
finally
does
the
customer
impact
and
for
us,
you
know,
being
Walmart
and
again
with
the
number
of
customers
that
we
serve.
This
is
huge
right.
So,
if
you
think
about
Lisa,
she
goes
to
the
store
thinking.
B
She's
gonna
be
ready
to
pick
up
her
grocery
she's
gonna
be
making
dinner
for
her
family,
and
if
we
disappoint
her,
you
know
there's
it's
tough
right.
It's
and
you're
really
affecting
everyday
people
when
things
don't
work,
so
obviously
quality
issues
impacting
our
customer
impacting
finance
financials
and
time,
and
something
that
you
know.
Clearly,
we
want
to
double
down
on
so
I'd
love
to
be
able
to
tell
you
that
you
know
we
deploy
perfect
code
all
the
time
and
we
have
no
bugs,
and
everything
is
good
and
Lisa
shows
up
and
she's
always
happy
unfortunate.
B
B
It's
that
you
develop
code,
you're,
also
very
much
responsible
for
the
quality
of
the
code,
and
our
job
is
also
to
make
sure
that
we
give
you
the
tools
that
make
it
easy
for
you
to
do
that
as
easy
and
as
pleasant
and
as
painless
as
possible.
So
what
do
we
do?
Certainly
from
you
know,
this
is
github,
so
I
will
say.
Github
is
a
very,
very
important
part
of
our
ecosystem
right.
You
know.
B
We
talked
about
the
scale
of
Walmart
we
talked
about
where
developers
are,
we
are
very
globally
distributed,
and
so
we
really
need
to
depend
and,
and
then
you
saw
the
services
and
how
they
depend
on
each
other,
how
well,
how
much
they
need
to
integrate.
We
need
to
make
sure
that
we
have
these
distributed
teams
code,
developing
services
and
we
need
to
depend
on
something
like
github.
Actually
not.
We
depend
on
github
to
provide
us
that
platform
for
doing
code
reviews
that's
very
important
to
us
from
a
culture
perspective
code.
B
Integration
is
also
the
ability
to
efficiently
share
and
leverage
code
across
teams
right,
so
I
think
one
of
the
things
that
we
talked
about
as
well
as
just
the
different
services
that
work.
We
also
have
different,
for
example,
of
different
ecommerce
sites
that
are
across
the
different
countries.
We've
got
store
systems,
obviously
in
the
US,
but
we
also
store
systems
in
other
parts
of
the
world
as
much
as
possible.
B
We
want
to
develop
services
that
are
multi-tenant
so
that
you
know
we
can
share
across
again,
just
as
all
around
efficiency
and
making
sure
that
we
can
do
that,
so
we
can
again
keep
costs
down
and
help
our
customers
out
as
well.
So
one
of
the
other
things
that
we
do
within
Walmart
labs
is
something
we
call
inner
sourcing
kind
of
a
term
that
we
coined
and
what
we
do
with.
That
is,
if
I
take
the
online
grocery
example
again,
so
you've
got
the
order.
B
Management
service
I
talked
about
how
it's
a
air
traffic
control,
it's
a
multi-tenant
service,
so
it
serves
online
grocery.
It
also
serves
the
Walmart
comm
app
as
well,
so
we
have
a
product
team
for
order
manager
order
the
OEMs
so
I'll
just
call
it
OMS.
We
have
a
product
team,
we've
got
developers
and
they're
really
focused
on
the
product
lifecycle
and
the
roadmap
for
the
OEMs,
obviously,
the
very,
very
critical
service,
they're
responsive
for
tech,
debt
and-
and
so
you
know,
this
Amos
lives
and
buries
OMS
right.
B
So
this
is
all
good,
but
then
we
might
also
have
feature
teams.
So
an
example.
The
cake
ordering
feature
team
say
that's
a
feature
we
want
to
introduce
in
walmart.com
and
they
need
something
to
be
done
in
OMS.
It's
you
know,
maybe
not
a
big
thing.
So
one
way
is
that
you
put
it
on
the
backlog
you
go
through.
You
know
these
prioritization
and
you
know
often
doesn't
get
done
because
it's
kind
of
a
you
know
not
the
biggest
most
impactful
feature
in
the
world,
but
for
customers.
That
is
a
delightful
feature
right.
B
So
what
we've
done
within
a
sauce
is
allowed
the
feature
team
for
cake
ordering
to
be
able
to
go
into
the
OMS
codebase
and
make
changes.
But
in
order
to
do
that,
we
need
to
have
again,
you
know
an
ecosystem,
a
platform
that's
going
to
allow
us
to
have
very,
very
extensive
documentation,
as
we
can
imagine
right.
You're
gonna
have
folks
that
are
coming
in
and
working
on
a
code
base
that
they
don't
normally
work
on.
B
So
documentation
has
to
be
there
again
the
code
review
process
and
the
tools
for
doing
that
have
to
be
spot-on
all
right.
So
we
need
to
be
able
to
do
active
code
reviews.
We
need
to
be
able
to
have
functional
integration,
testing
really
baked
in
to
the
process,
because
again,
you're
gonna
have
people
that
not
quite
as
familiar
with
the
code
base.
So
all
of
that
is
something
that
the
platform
has
allowed
us
to
do,
and
it's
sort
of
been
one
of
those
things
where
you
know
I
think
it's
it's
you
know
and
it's
hard.
B
Nothing
is
always
black
or
white
right.
It's
like
do.
You
always
have
feature
teams,
so
you
always
have
perfect
things.
Well,
reality
is
you
need
both,
and
so
something
like
inner
sourcing
on
a
platform
like
github
allows
us
to
sort
of
have
our
cake
and
eat
it
too.
So
again,
something
that
we've
been
we've
been
doing
pretty
effectively
and
really,
probably
something
we
need
to
do
more
off
across
Walmart
labs,
so
that,
like
I
said
it
only
works.
B
Did
I
jump
one
okay,
so
one
of
the
things
that
we're
working
on
as
well
is
to
ensure
that
we
make
it
easy
for
our
developers
to
adopt
this
culture
of
quality.
Is
that
we've
worked
on
a
platform?
It's
a
quality
automation
platform
called
test,
Armada,
it's
written
by
our
developers
for
our
developers
and
it's
written
on
top
of
a
lot
of
open
source
technology,
which
you
know
developers
in
general
of
the
years.
B
So,
let's
see
how
tesla
model
has
been
used
within
Walmart
today,
so
these
are
some
of
the
things
that
Tesla
Model
aoz
us
to
do.
You
know:
I
talked
about
functional
testing,
obviously
very,
very
important.
We
have
executors
over
19
million
test
cases
already
this
past
year
and
Tesla
Model
S.
With
that
there's
also
data
insights
platform.
It's
allowing
us
to
collect
all
the
test
data
from
disparate
sources
and
really
provide
insights
into
the
code
quality
code
coverage.
So
obviously
something
again.
That's
gonna
help
our
engineers,
debug
quality
issues,
there's
also
performance
testing.
B
So
now
this
is
again
something
that's
super
important
for
us
to
make
sure
that
we
include
in
sort
of
the
the
lifecycle
of
development
right
so
in
e-commerce
world.
One
of
the
things
we
call
conversion,
that's
essentially
when
you
come
to
our
site,
conversion
is
measured
when
you
actually
check
out
right
and
buy
something
from
the
site,
and
what
we
found
is
that
conversion
is
very
highly
dependent
on
how
quick
the
site
is
so
performance
clearly
is
one
of
a
very
positive.
You
know
if
you're
doing
well
with
performance.
B
It
has
a
very
positive
impact
on
your
what
we
call
gmv
or
your
conversion,
and
vice
versa
right.
So
one
of
the
things
we
want
to
make
sure
is
that
right
now
we're
we
have
a
big
project
that,
where
we're
looking
at
ensuring
that
we're
fixing
certain
things
around
performance,
but
I,
also
want
to
make
sure
that,
as
developers
are
continuing
to
add
features
to
the
codebase
that
it's
something
where
they're
not
actually
taking
us
back
right.
So
it's
not
like
you
know
one
step
back
one
step
forward,
two
step
back.
B
So,
in
order
to
do
that,
we
really
kind
of
baked
in
performance
testing
again
into
the
CI
CD
life
cycle,
and
then
the
last
thing
is
our
mocking.
So
again
you
know
you
go
back
to
that
example:
I
showed
you,
you
see
services
a
lot
of
interdependencies
and
that
was
actually
a
simplified
diagram
in
reality.
There's
more
interdependencies
than
that,
so
you
really
want
to
make
sure
in
terms
of
being
able
to
develop
and
not
be
dependent
on
another
service.
B
Being
a
bit
available
is
to
have
this
mocking
services
right,
so
it
helps
us
with
two
things.
It
allows
us
to
develop
more
quickly
and
tests
more
quickly,
even
when
the
other
system
is
not
available,
but
it
also
helps
us
detect
if
there's
something
that
we're
doing
some
code
that
we've
introduced.
Some
new
attributes
usually
are
the
things
that
kind
of
get
us
right
which
is
going
to
which
might
break
compatibility
with
the
service
that
you
services
that
you
depend
on
so
again.
B
The
mocking
the
mocking
API
is
something
that
we
depend
on,
probably
more
so
than
most
just
because
of
the
interdependency
of
the
services
that
we
have.
So
these
are
sort
of
the
the
different
things
that
Todd
tests
Hamada
helps
us
with.
You
know
clearly
very
important
to
our
CI
CV
lifecycle,
so
in
terms
of
putting
it
all
together,
yeah.
This
is
a
somewhat
simplified
view
of
our
CI
CD
pipeline.
We
talked
about
again,
you
know,
github
from
a
source
control
perspective,
helping
us
with
all
the
developers.
B
Do
you
have
globally
and
managing
source
code,
and
it's
all
good
and
then
part
of
the
CI
cycle
is
actually
making
sure
that
tesla
model
uggs
in
to
the
cycle
and
ensuring
that
every
change
actually
passes
a
quality
check
before
it
moves
along
its
merry
way.
Right,
cuz
I
think
it's
great
to
have
CI
CD,
but
if
you're
constantly
deploying
code
that
has
bugs
it's
actually
not
that
great
right.
B
B
So
we've
been
able
to
reduce
test
execution
time
from
hours
to
minutes
and
as
you're
doing
that
again,
you
know
you're
making
developers
much
more
productive,
you're
spending
less
time
testing,
but
you're
still
doing
all
the
testing
that
you
need
to
do.
So.
That's
that
fine
balance.
You
want
to
make
sure
that
you're
testing
the
heck
out
of
your
system,
but
you're
also
not
taking.
B
B
So
what
are
the
overall
benefits
of
tesla
model,
so
we've
got
around
85
projects
that
are
going
through
the
system
to
date,
and
this
executes
over
550
thousand
automated
test
runs
per
day,
which
is
about
30,000
engineers
working
per
day,
so
clearly
I
mean
the
silicon.
Imagine
like
thousands
of
combinations
that
you'd
have
to
test
right.
These
are
all
things
that
are
now
nicely
automated
and
if
you
can
think
about
that
again,
it's
around
you're
freeing
up
developers,
time
to
be
able
to
continue
to
innovate
and
build
better
products
for
our
customers.
B
So
for
those
of
you
who
are
building
platforms
and
tools
right,
it's
like
how
do
you
measure
how
good
your
tool
is
and,
and
how
do
you
measure
success
and
what
I
like
to
always
look
at
is
when
I'm.
Looking
at
my
my
teams
that
are
providing
platforms
and
tools
is,
how
is
it
used?
How
are
our
customers
adopting
it,
and
your
internal
customers
are
usually
your
most
difficult
ones
right
there
they're
kind
of
finicky
and
they
don't
mind
yelling
at
you
and
telling
you
what's
missing,
but
that's
that's
been
important.
B
It's
been
a
good
measure
of
success
in
terms
of
the
adoption
of
tests
Armada.
So,
as
you
can
see
many
groups
across
all
of
Walmart
that
are
using
it.
You
know
we're
in
various
locations
around
the
world,
and
so
it's
actually
being
used
very
globally
and
then
some
of
the
numbers
at
scaler
I
talked
about
the
projects
that
have
been
on-boarded.
B
There's
a
million
developer
hours
that
have
been
saved
by
using
test
ramada
17
to
19
million,
depending
on
how
you
count
it
functional
tests
executed
this
past
year
and
162
million
of
those
mocked
API
calls
that
I
said
you
know
again,
given
the
way
our
ecosystem
is.
Is
you
know
very,
very
important
for
us
to
be
able
to
do,
and
so,
where
are
we
going
in
the
future?
You
know
as
anything,
you
were
continuing
to
evolve
our
platforms.
B
So
a
couple
things
that
are
still
kind
of
in
research
that
we're
looking
at
you
know
one
is
sort
of
using
machine
learning
to
be
able
to
provide
a
means
for
automatically
generating
test
scripts,
so
reducing
time
spent
in
terms
of
creating
test
cases
and
maintenance,
so
something
that
we're
looking
at
right
now
in
a
research
perspective
and
hopefully
being
able
to
figure
out
how
we
roll
that
into
the
test.
Hamada
suite
we're
also
looking
at
automated
visual
testing
right.
B
So
we
want
to
be
able
to
detect
when
things
go
missing
without
actually
having
to
program
that
in
so
you
know,
this
is
a
use
case,
real
real
life
use
case.
You
know
one
at
one
point
simple,
CSS
issue
and
the
cart
and
checkout
button
has
disappeared
from
them
from
the
site.
Now
again,
if
you're
an
e-commerce
site
like
the
two
buttons
that
are
most
critical
to
you,
are
the
current
checkout
buttons
right
I
mean
those
have
to
be
there.
B
Everything
else
I
don't
care
about,
but
the
cart
and
checkout
button
has
to
be
there
so
again,
so
there's
a
drop
in
conversion
from
that
and
again,
if
there's
an
automated
way
that
we
could
detect
those
things,
it
would
be
helpful
for
us
as
well
and
then
finally,
a
predictive
and
analysis
right.
We
want
to
be
able
to
see.
If
we
can,
you
know,
obviously
you
wanna
be
able
to
run.
The
first
step
is
just
being
able
to
run
all
our
test
cases
and
making
sure
that's
exhaustive.
B
It
covers
code
and
it
covers
all
the
use
cases,
but
then,
over
time
you
want
to
see
we
can
optimize
right
and
just
make
sure
that
we're
picking
up
the
appropriate
test
cases
to
run
based
on
the
pull
request
compared
to
what's
in
history.
So
that's
something
that
we're
a
little
bit
further
along
and
we're
working
on
that
working
on
that
feature
actually
in
q4.
So
some
know,
that's
the
the
third
thing.
I
would
say
that
we're
doing
in
terms
of
helping
out
our
quality
automation
suite
so
that
was
tests
Armada.
You
know
it's.
B
It's
certainly
like
I
said
used
extensively
at
Walmart.
We've
got
some
external
companies
that
are
also
using
it.
The
nice
thing
about
open
source
is,
you
know.
Obviously
we
consume
as
much
as
we
can
and
we'd
like
to
be
able
to
give
back
to
the
open
source
community
as
well.
If
any
of
you
are
interested
in
finding
out
more
about
it,
please
check
us
out
online
and
then
I
would
say
you
know.
B
In
summary,
in
terms
of
what
we've
been
doing
at
Walmart,
certainly
like
I
said
you
know,
quality
is
is
a
is
a
big
concern
and
it's
something
that
something
very
important
to
us
anything.
It's
certainly
very
important
to
the
world's
largest
company,
but
it's
also
important
to
companies
that
a
little
bit
smaller
than
that
right
so
and
what
we
found
is
that
really
it's
around
developing
a
culture
around
quality,
where
it's
not
somebody
else's
promise
you
it's.
B
It
is
everybody's
shared
responsibility,
you
develop
code,
quality,
is
your
responsibility
and
in
terms
of
providing
tools
right,
that
was
the
other
thing.
I
would
say
it's
like
making
sure
that
we're
providing
tools
that
can
help
our
developers
encourage
this
culture
of
quality,
and
so
that's
what
we've
done
with
Tesla
Mata
and,
like
I
said
you
know,
we've
been
making
making
sure
that
the
usage
as
well
is
demonstrated
throughout
the
company,
so
also
I
would
I
would
say
you
know.
B
I,
hopefully
have
been
able
to
also
tell
you
a
little
bit
about
what
Walmart
labs
does
and
a
little
bit
of
how
tech
is
used
throughout
the
the
company.
We've
got
a
lot
of
interesting
tech.
You
know,
we've
we've
got
a
lot
of
machine
learning,
there's
robots
in
our
stores,
personalization
recommendations,
the
the
kind
of
usual
things
on
the
e-commerce
site,
and
so
you
know
I
would
encourage
you.
B
If
any
of
this
was
interesting
and
you
please
reach
out,
we
are
hiring
so
we'd
love
to
have
you
guys
join
and
you
know
finally,
I
would
say
thank
you
to
github
for
giving
me
the
opportunity
to
speak
to
you
all
today.
Not
only
did
I
get
to
present
I
got
a
whole
bunch
of
AI
steps
in
and
with
heels,
which
you
know,
I
think
raises
the
level
of
difficulty
so
I'm
feeling
pretty
proud
of
myself
right
now
so
anyways.
Thank
you.
Everybody.