►
Description
Join us for Kubernetes Forums Seoul, Sydney, Bengaluru and Delhi - learn more at kubecon.io
Don't miss KubeCon + CloudNativeCon 2020 events in Amsterdam March 30 - April 2, Shanghai July 28-30 and Boston November 17-20! Learn more at kubecon.io. The conference features presentations from developers and end users of Kubernetes, Prometheus, Envoy, and all of the other CNCF-hosted projects
Panel Discussion: Democratizing HPC & AI: Startups Scale Up with Cloud Native - Emily Tanaka-Delgado, Oracle; Charlie Davies, iGeolise; Priya Shah, Sauce; Ant Kennedy, Gapsquare; and Alfonso Santiago, ELEM
Innovativeness, resourcefulness, and responsiveness are all part of the startup DNA, and with the aid of the cloud native era and ecosystem, they are unstoppable. Hear how these European startups are leveraging technologies such as Kubernetes, containers, GPUs, Singularity and more to gain efficiencies across architecture, time, and team to reach global scale, reshape user interactions, and even, save lives. In this panel, startups from Barcelona and the UK will discuss how cloud native technologies have helped unlock access to unlimited potential across such diverse case studies as identifying and fixing gender pay gap issues, creating computational models of virtual patients, reinventing cloud-based video collaboration, and interpreting travel and navigation based on time, all to transform their businesses to match behavioral patterns within their industries.
https://sched.co/MPY0
A
Welcome
can
everybody
hear
me:
okay,
okay,
great
thanks.
So
much
for
joining
my
name
is
emily
tanaka
delgado.
I
lead
Deverell
engagement
and
oracle
cloud
and
I
am
joined
today
by
an
amazing
panel
of
startup
innovators
who
are
all
leveraging
and
taking
advantage
of
the
incredible
pace
of
development
and
access
to
cloud
and
open
source
technologies
before
we
dive
into
that
I'd
like
to
have
my
panelists,
please
introduce
yourselves
Charlie,
we'll
start
with
you.
If
you
could
tell
us
a
little
bit
about
your
company
mission
value
and
what
inspired
you
a
little
loud.
B
Hello,
great
Oh
got
even
louder,
hey
everybody.
My
name
is
Charlie
Davies
from
a
company
called
idealize.
We
have
developed
a
product
that
we
call
the
travel
time
platform
and
what
it
does
is
enables
the
searching
of
maps
and
data
in
the
world
around
us
by
time
instead
of
distance.
So
any
search
that
says
what
can
I
do
within
one
mile
five
miles
or
ten?
B
We
say:
what
can
you
do
within
half
an
hour
by
public
transport,
cycling,
walking
driving,
and
we
license
this
to
anyone
who
essentially
makes
money
through
locations
online,
so
real
estate
products
or
recruitment
products
or
hotels
searching?
We
now
do
this
in
about
35
different
countries,
and
if
you
want
to
play
around
on
it,
you
just
have
to
go
to
app
dot
travel
time.
Dot
me
on
your
phone
and
you
can
play
around
with
the
data
we
have
in
Barcelona
Thanks.
C
Kennedy
from
cap
square,
so
we're
a
b2b
startup
building
tools
for
companies
to
help
close
their
gender,
ethnicity,
disability,
etc.
Pay
gaps.
So
we're
running
running
on
our
cloud
to
enable
these
businesses
to
understand
how
they're
paying
their
employees,
if
they're,
paying
them
fairly.
So
research
from
Gartner
shown
future
workforce,
mainly
Millennials
within
20
year,
2020
care
more
about
transparency
and
fairness
with
their
within
the
company.
They
work
for
so
we're
looking
to
help
them
enable
this.
D
Hi
well
Alfonzo
from
Ellen
biotech
our
spin
office.
Actually,
it's
actually
a
spin-off
of
the
Barcelona
supercomputing
Center.
On
that
research
center
we
have
been
building
for
the
last
10
or
15
years
software
that
is
able
to
simulate
a
human
heart.
So
you
can
have
your
own
MRI
and
make
it
a
computational
model
of
that
I
hope
your
own
heart
beating
turn.
D
So
this
is
a
great
research
tool,
but
I
actually
want
to
put
this
tool
to
the
hands
of
people
that
can
make
an
impact
we've
done
right,
that
is
the
clinicians
and
the
engineers
that
make
that
design
the
biomedical
devices
that
goes
into
your
heart,
for
example
right.
So
we
have
to
program
with
this.
One
of
the
problems
is
that
these
simulations
are
really
complex,
complicated
to
to
set
up.
So
we
have
to
build
this
interface
that
allows
to
personalize
the
simulation
in
an
easy
way
making
transparent
for
the
user.
D
All
the
parameters
that
are
numerical
like,
for
example,
number
of
course,
or
where
it's
gonna
run
right
and
the
other
problem
that
we
had
that
we
are
tackling
with
the
cloud
is
the
supercomputing
power
right,
because
these
simulations
are
very
computationally,
expensive
and
not.
Everyone
has
a
supercomputer
under
their
desks
to
run
these
simulations.
So
we
are
using
cloud
to
put
these
super
computing
power
at
the
fingertips
of
the
people
that
can
actually
make
use
of
it
and
make
an
impact
with
this
a
tool
that
we
developed.
E
Priya,
hey
I'm
Priya,
one
of
the
cofounders
of
the
source,
video
and
we
are
a
cloud-based
video
creation
platform.
So
we
enable
brands
to
create
video
content
with
their
global
communities
and
powered
by
our
cloud
technology
and
smartphone
cameras.
So
businesses
need
to
create
more
video
than
ever
today
for
all
kinds
of
channels
and
reasons
and
typically
they've.
You
know
hired
media
agencies
to
produce
content
or
freelancers
switcher,
typically
expensive,
but
there
are
billions
of
people
that
have
a
smartphone
camera
device
on
them.
E
A
Thank
you,
and
as
part
of
this
now
hearing
everybody's
mission,
I'd
like
to
hear
a
little
bit
more
about
how
cloud
and
cloud
native
technologies
have
played
a
role
in
building
that
vision
for
you
and
I
know
I'll
phone.
So
you
already
dipped
a
little
bit
into
it.
But
maybe
you
can
expand
a
little
bit
more
for
us
sure.
B
So
we
kind
of
have
two
missions
or
two
goals
that
over
on
the
holes
that
accompany
that
the
first
one
is
to
remove
any
guesswork
from
any
location-based
search,
so
I'm
sure
we've
all
done
it,
but
you
do
a
search
for
any
local
content
and
you
see
a
map
and
the
map
gives
it
context,
doesn't
give
it
much
meaning
around
which
one's
best
for
you.
You
might
go
and
then
use
other
applications
from
the
application
you're
already
using
to
go
and
understand
that
further.
B
So
we're
leveraging
lots
of
transport
data
and
lots
of
localized
information
to
make
sure
that
when
you
do
those
location-based
searches,
you
don't
have
to
go
anywhere
else
from
that
site.
It's
all
there
upfront
ready
for
you
to
go,
and
the
other
one
is
to
get
more
people
to
use
public
transportation
or
other
motor
transport
in
urban
areas.
It's
essentially
remove
the
need
or
use
for
any
emission
producing
vehicles.
B
We
know
that
eventually,
electric
vehicles
will
solve
a
lot
of
that
problem,
but
there
are
problems
today
that
we
can
solve
in
terms
of
moving
people
around
more
efficiently
using
public
transport
data,
especially
in
the
enterprise
market,
in
the
countries
that
we're
that
were
in
and
in
the
countries
element
that
that
there
provides
one
of
the
biggest
problems
that
we're
trying
to
solve
on
mass.
Is
this
huge
amount
of
data
that
we
have
to
take
in
for
all
these
places,
so
we're
not
just
City
centric?
B
We
have
to
map
the
entire
country
for
our
solution
to
be
as
useful
as
it
can
be,
and
also
cross
border
countries
as
well
and
public
transport
data.
You
may
think
is
really
static.
You
know
it
just
stays
the
same,
but
it
doesn't.
It
changes
a
lot.
I,
never
thought
that
I'd
become
a
digital
train
spotter,
but
that
is
something
that
I
have
actually
become,
and
we
take
all
this
data
and
we're
having
a
problem.
Essentially
the
machines
we
were
using.
B
You
know
our
code
base
grew
up
very
attached
to
the
the
revenue
we
were
getting
from
our
early
clients,
I'm
sure.
We've
all
been
there
when
you
release
the
product,
clients
are
using
it
your
conditions
around
getting
code
out
change
and
we
had
to
move
our
infrastructure
back
on
to
now
running
kubernetes
in
the
cloud
to
change
our
map
processing
time
down
from
that.
If
it
was
up
to
two
weeks
now
down
to
a
few
hours,
so
we
can
now
take
that
data
refresh.
B
It
make
sure
it's
back
out
there
for
our
users
as
well
as
use
that
for
testing
as
well.
So
the
democratization
element
of
actually
allowing
individuals
in
our
team
who
aren't
that
proficient
in
encoding
or
or
technical
technical
skills
can
go
and
take
data.
Spinning
up
on
the
road
instances,
take
the
tools,
you've
already
developed
test
that
against
the
environment.
We
have
and
then
eventually
deploy
that
on
their
own
instances
into
our
production
environment.
That's.
A
D
Of
course
yeah,
so,
as
I
told
you
before,
these
simulations
are
real
expensive
right.
So
when
you
are
in
a
research
project
in
a
research
center
that
works
for
the
public
administration,
they
say:
that's
that's
pretty
sitting
cycle.
That's
that's
not
a
problem,
because
you
have
a
supercomputer
that
it's
a
homogeneous
cluster.
You
don't
have
to
spend
money
on
this
right,
so
yeah.
D
The
accessibility
to
this
super
computing
power
is
like
easier
because
you're,
a
part
of
this
right
now,
when
you're
trying
to
move
when
you're
trying
to
make
an
application
a
useful
application
from
this
research
tool.
It's
you
have
to
solve
that
problem.
You
need
to
you
need
to
to
understand.
How
do
you
have
you
make
this
super
power
super
computing
power
available
for
for
the
industry,
and
these
type
of
simulations
are
completely
computationally
expensive.
What
I'm
this
means
is
that
we
generally
for
us
for
a
generic
run.
D
We
use
like
about
200
cores
about
five
hours,
three
hours,
depending
on
the
case
right.
So
when
you,
you
need
to
run
this
outside
of
the
computer,
and
the
cloud
is
the
perfect
target
for
this,
because
it's
super
computing
power,
you
can
absolute
super
computing
power
that
is
available
without
actually
having
the
computer
under
your
desktop
or
in
the
or
in
your
place.
D
You
only
pay
when
you
use
it,
so
it's
much
more
plug
practical
and
flexible
for
an
applet
for
for
a
business
like
this
right,
so
the
user
of
this
software
is
the
one
that
is
gonna
choose
when
it's
gonna
have
this
little
bit
of
power
in
which
way
how
its
gonna
pay
for
that.
So
that's
really
interesting,
and
also
it's
much
more
flexible
than
actually
piece
of
the
computer
that
we
have
on
the
research
center.
Why?
D
Because
it's
allows
you
to
run
different
schemes
of
different
execution
frameworks,
for
example,
and
we
can
run
for
larger
cases.
We
tend
to
run
like
per
metal
instances,
but
we
something
else
that
would
you
is
something
that
we
call
a
virtual
populations,
that's
something
where
we
try
different
configurations
on
different
hearts.
It's
like
populations
or
also
it's
also
something
like
a
sensitivity
analysis.
So,
in
those
cases
what
we
do
is
we
run
the
executions.
We
run
the
execution
with
containers
because
it's
much
more
flexible.
D
It
is
completely
independent
on
the
infrastructure
or
the
architecture
of
the
computer
you're
running.
So
the
cloud
basically
made
this
project
possible
and
what
makes
is
being
able
to
anyone
that
is
interested
in
running
and
having
the
results
of
such
simulations
to
have
the
simulations
in
a
short
time
and
being
this
power
available
for
anyone
and.
A
E
Absolutely
so
you
know
our
vision
at
source
is
to
really
empower
businesses
to
create
with
their
communities,
so
typically
the
kinds
of
users
that
use
our
apps
and
platform
and
not
traditional
videographers
they're
everyday
people
right.
So
you
know
we
use
a
lot
of
that
technology
to
make
that
a
whole
process
of
creating
a
video
really
really
easy.
And
so
you
know,
over
the
past
few
years
we've
been
capturing
a
lot
of
data
about
how
our
users
are
editing
their
video
on
their
mobile
phone.
So
that's
capturing
data
around
where
they
trim
their
videos.
E
What
kinds
of
music
and
filters
they
add
where
they
might
slo-mo.
You
know
particular
parts
and
we're
now
using
that
data
to
actually
assist
editing.
We
don't
like
to
call
it
auto
everything.
It's
more
assisted
editing,
so
we
can.
Actually,
you
know,
do
some
of
the
heavy
lifting
when
it
comes
to
editing
a
video,
so
I
don't
know
if
any
of
you
guys
have
ever
editing
a
video
before,
but
you
know
it
can
take
hours
if
you're
sitting
at
your
desktop,
you
know
tweaking
clips,
but
actually
with
our
technology.
A
C
So
we
started
out
and
fairly
simple
running
everything
on
a
single
VM,
deploying
binaries
ourselves,
and
that
was
great.
The
product
was
very
simple
at
time.
Doing
some
basic
analysis,
a
company's
HR
data,
as
as
the
company's
grain
and
as
our
clients
we
get
some
bigger,
will
be
doing
more
complicated
analysis,
so
we're
able
to
now
start
to
work
out
which
elements
of
your
business
are
contributing
to
your
pay
gap.
This
obviously
takes
a
little
bit
longer
to
compute
and
leads
to
a
less
than
ideal
user
experience
using
the
applicator.
C
We're
also
introducing
insights
into
actually
explaining
where
you
might
have
unexplained
pay
gaps
within
your
company,
so
say,
for
example,
there's
a
difference
between
men
and
women
using
our
machine
learning,
we
can
contribute
how
much
pay
a
women
and
how
much
pay
a
man
earns,
and
it's
there
any
unexplained
gap
in
that.
So
this,
the
moved,
cart
native,
has
enabled
us
to
grow
and
scale
the
platform
massively.
C
So
us
trying
to
run
that
all
on
a
single
VM
running
that
analysis
alongside
the
existing
application,
alongside
actually
deploying
a
new
machine
learning
motives,
new
versions
of
our
software
would
have
been
an
absolute
nightmare
using
platforms
like
kubernetes.
We
can
actually
start
to
a
B
test.
These
components,
deploy,
deploy
new
models
in
a
reasonably
seamless
manner
and
enable
these
insights
to
our
users.
So.
C
Open-Source
is
absolutely
fantastic,
I'm,
generally
quite
a
lazy
person
somewhere,
it
exists.
Let's
go
ahead
and
use
it,
it's
not
always
fit
for
purpose.
That's
fine!
You
should
be
building
on
the
shoulders
of
giants
and
particularly
with
the
cloud
native
community,
absolutely
amazing,
on.
If
you
go
to
any
of
the
projects,
people
are
more
than
willing
to
help,
which
is
a
huge
part
in
getting
any
of
this
done.
So
so
we're
not
back.
In
the
day
when
we're
running
on
VM
console
logs
that
work
log
into
files,
that
was
it
that's
a
production
goes
down.
C
We
get
the
CEO
gets
an
email
from
the
customer.
We
get
an
angry
email
from
the
CEO.
That's
that's
less
than
ideal
using
projects
like
Prometheus
flu
Andy,
we
in
30
30
minutes.
If
that
we
can
have
logging,
aggregated
logging,
Metrix
collection
alerting
say
that
the
tech
team
now
knows
that
there's
a
problem
before
the
CA,
so
we
can
manage
it
better
as
it.
A
C
A
C
We
we've
had
an
from
a
from
a
team
side,
an
interesting
journey
say
every
everybody
quite
likes
to
build
their
own
thing
and
take
ownership
of
that
so
using
open
source.
We
have
taken
a
few
of
these
projects
and
little
contributions
back
ourselves,
so
helping
give
back
to
the
community.
But
in
terms
of
our
processes,
we
we
can
now
call
the
juice.
C
We
we're
now
moving
to
more
of
a
end
to
end
ownership.
Ourselves
say
keeping
a
rapid
deployment
like
like
these
open
source
projects,
Kuban
at
Eve's,
fair,
fairly,
regular
release
cycle,
sometimes
difficult
to
keep
up,
but
we've
taken
this
mentality
and
housed,
say
we
we've
moved
to
continuous
deployment,
say
we're
getting
our
software
updates
out
to
our
customers
as
quickly
as
possible
and
and
hopefully
they're
enjoying
the
benefits
sooner
rather
than
perhaps
in
more
delayed
release
cycle
that
we
were
using
previously.
C
A
D
Of
the
workers
yeah
open
source
software
is
like
transversal
to
an
absolutely
necessary
in
every
research
area.
When
you're
doing
research
right
because
iran
and
linux
busters,
you
use
open
service
solution
tools.
You
use
a
lot
of
open
source
tools
to
do
that.
So
for
our
for
us
that
natural
environment
has
like
moved
naturally
to
the
company,
we
are
in
so
open
source.
Software
is
like
everywhere,
probably
like
90%
of
the
software
that
we
use
is
open
source
right,
even
ours,
ours,
so
our
own
source
code
is
open
source.
D
So
it's
transversal
and
I
mean
this
type
of
this
type
of
way
of
working
has
also
it's
also
in
everyone
on
the
company
and
on
the
research,
and
it
is
very
natural
sharing
your
work,
helping
others
to
do
something
better
with
the
tool
you
do.
You
got
right,
and
that
makes
you
that
makes
you
proud
like
helping
someone
else,
it's
something.
Yet
you
have
done.
That's
really
nice
and
that's
that's.
Open-Source
mentality
is
like
in
everyone,
India,
Research,
Center
and
in
the
company.
So
that's
the
idea,
helping
others
right
with
the
things
you
do
absolutely.
A
And
in
your
case,
trying
to
use
the
tool
you're
using
open-source
but
you're
also
providing
a
tool
that
you're
guiding
your
users
so
that
they
can
access
it
and
leverage
the
power
of
it.
But
maybe
not
know
every
single
thing
about
it.
Right
not
go
through
the
research
that
you've
already
done,
but
really
help
to
curate
that
for
them
to
be
able
to
do
what
they
need
to
do.
Yeah.
D
In
in,
in
our
case,
the
final
tool
that
we
are
doing,
it's
the
thing
with
the
thing
with
open
source
is
that
you
sometimes
need
the
sirs.
Not
everyone
is
easy
services.
Someone
some
people
just
need
yep
executable
file,
the
compiler
right,
but
I
mean
some
cases
they
do,
because
they
you
want
to
know
what
you
think
and
they
want
to
maybe
make
their
modifications
of
that
software
to
make
it
fit
it
to
their
applications.
So
that's
what
we're
targeting
I
mean.
D
E
I
think
I'm,
one
of
the
things
that
which
strikes
me
as
when
we
bring
customers
and
clients
to
our
office,
to
you
know,
for
a
meeting
or
to
meet
with
us.
They're,
really
surprised
that
we're
a
team
of
eight
people
Wow.
How
do
you
build
all
of
this?
How
do
you
manage
this?
You
know
we
have
global
clients
like
Accenture
and
Oracle,
and
Premier
League
football
teams
and
they're
amazed
and
I
think
you
know,
because
we
have
access
to
use
all
these
technologies
that
are
in
the
cloud.
E
We
can
manage
our
customers
and
we
can
manage
heavy
loads
and
videos
that
are
getting
uploaded
to
our
to
our
platform,
and-
and
you
know,
we
can
really
focus
on
building
incredible
kind
of
technology,
for
our
customers,
for
the
end
user,
and
so
taking
that
you
know
that
strain
and
our
development
team,
and
they
can
really
focus
then
on
building
features,
and
you
know
things
our
customers
really
want
to
use
and
improving
our
platform
on
that
side.
So
yeah.
A
A
Great
and
I
think
the
extension
value
there
is
huge
for
open
source
I
mean
this
conference
is
a
testament
to
that
right.
It's
tons
and
tons
of
people,
and
that
means
that
we
all
get
basically
that
amplification,
just
by
participating
and
talking
to
other
people
learning
what
they're
doing
learning
what
went
wrong
right
learning
what
the
surprises
were,
and
that
really
helps
us
all
move
forward
at
a
much
faster
pace,
which
is
part
the
whole
benefit
right
and
I.
Think
that's
a
you
know,
kind
of
a
nice
segue
just
to
kind
of
dip
into.
A
B
Our
open,
so
we
you
know
I,
think
the
any
of
us
would
be
here
in
this
open
source
really
existed.
We
use
open
source
in
a
number
of
different
ways,
whether
it's
down
to
under
the
libraries
we're
using
geospatial
technologies
or
even
down
to
the
data
that
we
take
in
remote
data,
be
used,
for
example,
OpenStreetMap
is,
is
it
freely
available
to
use
even
commercially,
but
it's
not
always
completely
correct.
B
So
we
do
a
lot
of
work
by
taking
that
data
cleaning
it
and
then
giving
it
back
straight
to
the
community
or
even
the
public
transport
agencies
that
we
work
with
the
one
time
that
I've
got
in
my
mind,
about
open
source
being
a
big
problem
was
when
we
sat
down
with
a
lawyer
from
a
bank.
We
were
deploying
our
technology
with
it
with
a
bank
in
the
UK,
and
they
just
didn't
get
what
open
source
was.
They
wanted
to
read
every
single
license
through
and
through
and
through
one
of
them
said.
B
If
you
ever
meet
me,
you
can
buy
me
a
beer
and
they
were
worried
that
if
they
used
it
that
many
times
how
many
beers
they
would
actually
have
to
buy
the
guy
who
wrote
the
license
on
the
readme
file
on
that
github
repository
so
sometimes
is
more.
The
Education
around
other
people,
not
understanding
completely
what
open
source
is
and
how
it
works,
especially
in
the
commercial
environment.
B
When
people
really
it's
their
job,
to
make
sure
that
you
know,
especially
for
a
bank,
had
to
tie
everything
down
and
also
you
know,
just
cuz,
it's
open
source
and
free
doesn't
mean
it's
completely
the
right
decision.
You
know
we
have
to
make
sure
that
we're
using
tools
that
accelerate
what
we're
trying
to
do
right
now
when
we
started
there
was
a
big
thing
around
I
when
I
was
when
I
was
running.
I
was
actually
coding
everything
it
was.
B
How
can
we
do
everything
without
spending
any
money
at
all,
and
there
were
lots
of
open
source
libraries.
We
could
use
some
of
them.
We
use
now
actually
proprietary
because
they
are
accelerating
what
we
can
do
ahead
of
time.
But
now
we
all
of
our
developers,
spend
20%
of
their
time
contributing
back
to
open
source
projects,
whether
that's
on
the
data
side
or
the
programming
language
we
use
or
right
down
to
the
library
to
be
used
as
well.
A
That's
great
what
about
total
impact?
You
know
for
everything.
We've
talked
about
from
open
source
to
foul
to
fab
native
technologies.
What
do
you
see
across
impact
for
your
company,
your
customers?
Even
your
team
right,
we've
talked
a
little
bit
about
it,
but
maybe
some
of
you-
and
maybe
there
are
some
things
you'd
like
to
expand
on,
say.
C
In
terms
of
what
cloud
actually
offers
the
gap,
Square
is
a
platform
where
we
are
able
to
start
relying
on
other
people
to
do
the
work
for
us.
So
there's
many
managed
services
out
there
that
we're
making
great
use
of
and
avoiding
doing
the
hard
things
ourselves
so
we've.
If
we,
if
we
weren't
able
to
rely
on
those,
we
would
have
to
you,
bring
that
that
talent
in-house
say,
spend
more
money
on
recruitment
and
less
money
building
features
until
more
money,
managing
updates,
etc.
And
it's
it's.
C
C
A
D
I
also
answer
questions.
You
may
need
Charlie,
okay,
okay,
so
we
we
had
like
two
peak
aha
moments
that
like
change
a
lot,
the
input
we're
having
the
philosophy
behind
those
aha
moments
is
like
clear
and
nobody.
Everybody
will
say.
Yes,
of
course,
that's
true
I
mean
I
mean,
but
it's
very
really
very
hard
to
understand
what
you're
doing
right.
So
one
of
the
moments
is
that
we
work
with
results
that
should
be
understood
by
clinicians
and
physicians
right.
D
You
won't
get
it
I
mean
the
same
reasons
we
show
to
them
in
the
slices
that
we
were
used
to
see
and
the
sequence
and
the
organization
of
the
image
that
they
were
used
to
see,
and
in
that
moment
they
started
saying
oh
right,
a
human
heart,
yeah
I
see
how
that
it
moves
I,
see
the
different,
the
different
strains
and
see
AppData,
you
start
seeing
everything.
So
what
is
the
story
behind
this?
D
It
is
that
we
didn't
understood
that
we
had
to
feed
the
client
according
the
what
I
wanted
to
be
filled
right
now,
so
they
we
were
trying
to
feed
them.
What
we
are
used
to
to
get
from.
So
this
you
can
extrapolate
this
everything
you
cannot
like
put
a
dish
in
front
of
the
of
the
client
saying
hey.
It
is
because
I
met
it
right.
You
have
to
sell
that
thing
in
a
way
that
they
feel
appealing
right.
D
We
need
to
do
something
that
these
people
can
use
right.
We
cannot
do
that
India
and
the
research
center
on
for
the
state
right,
because
that's
not
something.
That's
not
research
that
is
gonna
lead
to
increase
of
knowledge.
That's
research
or
work
that
it's
gonna
lead
to
increased
a
welfare
of
the
world
by
making
this
tool
available
for
who
needs
that
right.
But
I'm
in
that
moment
is
when
we
said.
Okay,
how
do
we
do
is
well,
we
have
to
create
a
company.
So
that's
the
other
moment
we
had.
A
E
E
Sometimes
we
surprised
ourselves
with
how
much
you
know
how
much
kubernetes
and
things
can
help
us
scale
at
those
critical
moments
where
you
think
it
might
fail,
but
it
actually
helps-
and
you
know,
we've
had
several
of
those
moments
where
we've
taken
on
stuff
and
it's
it's
worked
and
I.
Think
that
has
just
been
really
positive
and
we,
you
know
we
embrace
technology
even
more
and
especially
cloud
technology
when
those
moments
happen
and
you
you
come
back
and
you're
like
what
are
this
thing?
Like
really
working
work,
which.
A
Is
which
is
a
great
moment
to
really
something
works
right,
we're
pretty
happy
when
that
happens,
yeah
and
and
in
fact
some
of
these
things
really
just
couldn't
have
happened.
You
know
even
a
few
years
ago
in
terms
of
the
pace
of
Technology
and
we're
even
cloud
was
a
few
years
ago
and
and
I
think
those
those
moments
of
scale
Charlie
I
know
you
had
some
of
those
those
moments
of
trying
to
scope.
B
B
B
Because
the
system
was
really
bad,
I
built
it.
Luckily,
I've
got
far
more
intelligent
people
working
for
us
now
and
then
recently
we
launched
a
service
that
is
able
to
take
a
hundred
thousand
locations,
rankin
sort,
those
and
deliver
them
back
to
the
client
in
under
250
milliseconds,
including
the
network
transfer
time.
B
Actually,
we
know
very
well
when
that
limit
is
going
to
get
reached
in
our
current
architecture.
If
we
reach
that
point,
maybe
we
won't
care,
but
we
do
know
at
the
moment.
The
designs
that
we
put
in
do
have
limitations
in
terms
of
the
scale
the
concurrency
rates,
the
overall
trailer
network
capabilities.
B
C
Yep,
so
we
we're
very
much
looking
forward
to
where
we
can
take
the
machine
learning
side
of
our
business.
So,
given
the
availability
and
the
democratization,
the
democratization
of
AI
there's
so
many
open-source
projects
out
there,
that
will
do
these
cool
little
jobs
for
you
and
it's
now,
given
the
state
of
containerization,
it's
very
easiest
for
us
to
go,
build
a
container,
deploy
it
stick
an
API
in
front
and
great
we've
got
a
new
feature
in
the
tool
that
we
can
start
selling
to
customers.
C
I
admit
it's
not
quite
as
easy
as
that,
but
we
we
can
very
much
sell
it.
A
build
proof
of
concepts.
A
lot
quicker
and
refining
takes
a
lot
longer,
but
it's
again
this
this
would
not
have
been
possible
even
five
years
ago.
We
look
at
the
stay
of
machine
learning
back
then,
and
it's
well
then
so
this
machine,
yeah
state
machine,
then
back
then,
and
the
state
machine
now
the
amount
of
compute
resources,
particularly
techniques.
It's
it's
way
outside
this
gate.
What
any
small
tiny
company
wouldn't
be
able
to
do
back
then.
D
Yeah
for
us,
the
tip.
The
main
reason
that
made
us
this
company
possible
is
was
yeah,
nothing
much
the
computer
power
because
we
had
the
computer
power,
but
the
possibility
to
have
the
super
computing
power
available
for
anyone
right.
Thanks
to
that,
we
can
build
a
company
that
can
make
a
software
that
can
be
used
by
anyone,
because
our
application
requires
super
computing
power
that
wasn't
possible
before
that
not
five
years
ago.
D
D
But
at
some
point
we
would
like
to
move
towards
clinicians
and
imagine
this
future
close
near
future
that
you
go
to
the
clinician
and
you
go
there
and
he
gets
your
MRI
right
and
you
have
an
infant
and
you
can
put
that
MRI
on
the
software
and
you
get
a
simulation
of
your
in
fact,
a
hub
heart
right
and
the
clinician
can
try
on
the
computer.
What
is
the
best
treatment
for
you?
Where
is
the
best
pacemaker?
What
is
the
best
location
for
a
pacemaker
before
you
get
even
in
the
surgery
room?
D
And
then,
when
you
move
to
the
sensory
room,
you
already
know
what
is
the
most
efficient
treatment
for
you
right,
because
this
is
another
problem.
We
all
think
I'm.
At
least
I
thought
that
if
you
get
a
pacemaker,
it's
gonna
work,
I
mean
everybody
that
gets
a
pacemaker,
thinks
it
the
thing
that
he
has
inside
and
it's
gonna
work
right.
That's
not
the
case
and
every
time
right
today,
people
that
have
this
are
very
specific
terms,
but
people
that
has
left
ventricle
branch
block.
D
Forty
percent
of
forty
percent
of
the
patients
that
get
a
pacemaker
they
pacemaker
is
useless.
It's
not
working
properly
because
at
the
pathology
is
not
responding
to
the
pacemaker.
So
with
this
type
of
software,
what
we
wanna
do
is
improve
these
rights,
get
a
pacemaker.
If
you
need
a
pacemaker,
get
the
most
optimal
configuration
of
your
pacemaker,
for
you
get
the
better
the
best
pacemaker
for
your
pathology.
So
that's
what
we
aim
in
10
years.
It's.
E
Yeah
and
you
know,
I
think
putting
my
sales
and
marketing
hat
1.
For
me,
it's
all
about
creating
a
great
product
that
people
want
to
use.
Ultimately,
people
need
to
use.
You
know
the
technology
that
you're
developing
for
you
to
be
a
viable
business
right,
and
so
these
technologies
are
great
and
there's
a
lot
out
there.
But
at
the
end
of
the
day,
it's
it's
about.
Knowing
what
you
need,
because
there's
lots
you
can
build,
there's
lots
of
tools
available,
and
but
it's
really
about
for
us
as
a
team
and
I.
E
E
So
you
know
I
think
it
all
comes
down
to
you
kind
of
working
together
with
product
teams,
with
sales
teams
and
understanding
the
pure
use
case,
and
why
are
we
building
this
like?
Will
it
make
our
technology
faster
or
what
is
it
that
you're
building
so
I
think
that's
pretty
important
and
we
kind
of
really
believe
in
that
I'm
working
together
as
a
team
and
when
we're
talking
about
these
technologies
and
choosing
what
to
use
great.