►
From YouTube: CDF - SIG MLOps Meeting 2021-08-12 (1 of 2)
Description
For more Continuous Delivery Foundation content, check out our blog: https://cd.foundation/blog/
A
A
A
B
B
B
B
I
I'm
not
hearing
you.
B
Welcome
so
have
you
been
to
one
of
these
sessions
before.
A
No
and
unfortunately,
I
just
noticed
that
I
have
to
go
to
a
shop
and
get
my
bike,
so
I
I'm
actually.
This
is
the
first
time
I've
been
to
that
one
of
those
meetings,
and
I
have
noticed
that
I
cannot
show
the
tent
because
of
I
need
to
go.
I
need
to
fetch
my
bicycle.
B
Okay,
no
problem
I'll
we'll
we
look
forward
to
seeing
you
another
event
anyway,.
A
But
yeah
it
looks,
looks
quite
interesting.
I
think
I
tried
to
join
one
of
those
meetings
like
a
few
months
ago,
but
I'm
not
sure
if
there
was
the
link
still
active
or
I
was
not
sure
if
this
still
was
taking
place,
but.
B
Yeah
yeah,
so
the
the
the
meetings
are
generally
every
two
weeks:
okay,
on
this
time
slot
and
then
once
a
month
on
the
asia,
pacific,
timeslot,
yeah
cool
and
we're
we're
a
bit
quiet
at
the
moment,
because
it's
summer
holidays
and
everybody's
away,
that's
right.
B
I
I'm
actually
based
in
jersey
in
the
channel
islands,
all.
B
So
did
you
want
to
just
quickly,
let
me
know
what
your
interests
are
and
how
I
might
be
able
to
help.
A
So
yeah,
I'm
not
I'm
I
as
trustworthy,
I'm
going
to
join
the
meetings,
it's
all
like
ml,
ops,
related
right,
and
so
I'm
like
I'm
working
in
a
ai
department
of
our
company
and
I'm
the
like
taking
care
of
like
bringing
stuff
into
production
like
I'm,
not
a
data
scientist,
but
I'm
helping
to
bring
stuff
into
production
so
yeah
and
I
think
all
this
ml
top
ml
ops
related
things
are
cool
yeah,
so
you're
definitely.
B
So
what
we
are
primarily
focused
on
in
this
group
at
the
moment
is
the
2021
update
to
the
ml,
opt
road
map.
Okay,
have
you
seen
the
roadmap
document?
I.
A
I
just
was
that
the
landscape-
I'm
not
sh
it
was-
is
this
from
the
lf
just
data
foundation,
interactive
landscape:
do
you
mean
that
one
or.
A
A
Perfect,
I
bookmarked
it
and
read
it
later:
yeah.
B
Yeah,
so
if
you
have
a
flip
through
that,
that
will
give
you
a
good
idea
of
what
we're
doing,
which
is
really
to
try
and
provide
one
location
where
you
can
find
all
of
the
potential
challenges
that
you
might
hit.
If
you're
working
with
machine
learning
assets
in
production
environments
and
then
an
analysis
of
some
of
the
technology
requirements
that
come
out
of
that,
and
then
a
map
of
what
which
of
those
challenges
are
already
very
well
addressed,
which
is
verses,
which
are,
you
know,
currently
unsolved
and
likely
to
cause.
You
problems.
A
Cool,
I
definitely
need
to
forward
this,
so
I'm
like
we
have
a
student
here
that
will
soon
write
his
master
thesis
in
the
for
the
ml
in
the
ml
ops
ecosystem
and
to
to
figure
out
what
patterns
are
there
and
so
on.
So
I
definitely
need
to
forward
that
to
them
that
to
him
yeah
and
cool-
and
I
also
will
have
a
read
here-
looks
so
it's
quite
a
big
one.
Actually.
B
So
I
I
work
as
a
consultant
cto
helping
people
to
solve.
You
know
these
types
of
problems
at
scale,
okay,
and
so
I
get
involved
with
when
companies
are
trying
to
set
up
doing
this.
For
the
first
time,
then
we're
going
to
help
them
out
and
point
them
in
the
right
direction
and
help
them
build
their
teams
and
set
up
best
practices
and.
A
Yeah,
that's
it's
quite
interesting
yeah.
We
I
mean
we
are
facing.
Also
these
issues
currently
are
not
like.
We
are
building
our
applications
and
yeah
the
the
we
are
not
using
kubernetes
at
that
point,
but
we're
using
the
nomad
stack,
which
is
also
quite
interesting
because
you
don't
have
the
the
same,
like
I
think,
for
a
kubernetes
deck.
You
have
a
like.
All
these
technologies
are
ready-made
technologies
for
the
kubernetes
deck,
but
in
the
nomad
world
it's
a
bit.
B
B
B
And
so
using
containerization
and
standard
platforms,
and
things
like
that
makes
life
easier,
but
the
the
mlop
support
in
that
area
is
still
quite
weak.
B
We
hopefully
will
continue
to
improve
that
and
simplify
the
way
that
that
works.
A
I
haven't
used
jenkins
in
a
like
for
real,
like
at
the
moment.
I'm
I'm
stuck
with
that
azure
devops
thing,
but
yeah
I
mean
jenkins
x,
looks
generally
very
interesting,
but
also
it's
like
one
of
those
technology
you
can
use.
If
you
have
a
kubernetes
stack
and
yeah.
As
I
said,
we
we
have
this.
We
we
use
nomad
where
it's
like.
There
is
nothing
real.
I
mean
you
can
do
similar
things
like
with
kubernetes.
B
A
That's
true
yeah.
B
It
can
be
daunting
to
start
with,
but
actually
it's
the
same
pattern
everywhere.
Yeah
yeah.
A
B
You
understand
what
the
pattern
is.
You
can
apply
that
to
every
single
problem
that
you
need
to
solve
in
that
space
and
it
will
always
work
in
the
same
way.
So
it
reduces
your
over
overall
complexity
load.
Quite
a
lot
and
and
jenkins
x
hides
a
lot
of
the
nastier
stuff
so
that
you,
you
don't
need
to
spend
a
lot
of
time
manually,
generating
yaml
and
pushing
things
into
production.
B
A
Yeah
yeah,
I
mean
I
used
kubernetes
before
other
job,
but
yeah.
No,
I
mean
this
whole,
you,
you
did
the
nomad
the
hashicorp
thing
you
you
know
that
one
or
is
it.
A
It
I
think
it's
pretty
interested
like
I
think
they
have
like
some
stuff,
where
we
actually
think
it's
it's
nicer
to
have
it
like
the
hcl
configuration
language,
I
think,
is
more
like
you
don't
need
that
yaml!
You
don't
have
that
yamaha
and
like
it.
It
organizes
quite
nice
with
with
terraform
and
devold
and
with
volt
and
all
that
stuff.
So
I
think
it's
also
quite
nice
technology.
I
think
here
they
start
using
it
because
they
they
decide
it's
it's
easier
to
maintain
the
the
volt
this.
This
had
this.
B
Yeah,
I
think
it
very
much
depends
what
what
the
nature
and
scale
of
your
problem
is
yeah,
because
there
are
many
things
where
given
that
this
is
just
overkill
for
that.
But
if
you're,
if
you're,
trying
to
build
things
that
are
need
to
scale
very
large
and
need
to
be
very,
very
resilient,
then
it's
hard
to
beat
that
yeah.
A
Definitely
no,
I
mean
I
like
it,
but
it's
at
the
moment
it's
what
I
have
to
work
with.
So
it's.
I
cannot
go
to
the
ops
department
and
say
please
change
that
and
yeah,
but
I
see
it
as
a
challenge
and
I
think
I
hopefully
maybe
I
have
this
this.
You
know
you
have
like
these
kind
of.
There
are
a
lot
of
claims
to
take
in
the
in
the
the
nomad
ecosystem
that
are
like.
There
are
many
of
these
technologies
available
for
the
kubernetes
ecosystem.
A
Like
there
is,
I
don't
know,
k
native.
There
is
k,
there
is
all
the
egg
or
there
is
the
like.
All
the
technologies
they're
like
running
natively
on
on
kubernetes,
especially
in
ml
ops,
world.
I
think
yeah.
I
see
this
as
a
challenge
that
we
can
maybe
create
create
some
nice
nomads
native
ml
ops
related
technologies,
but
yeah
that
would
be
nice.
B
Yeah
yeah:
that's
something
that
we're
trying
to
encourage
in
in
this
forum.
You
know
what
we
really
want
to
do
is
get
people
to
focus
on
what
the
customer
problems
are
in
terms
of
mrs
and
then
start
to
factor
that
into
the
existing
cicd.
A
A
So
sorry-
and
I
noticed
I
need
to
put
my
monitor
like
your
screen
on
the
same
screen
as
the
the
where
my
webcam
is
so
then
it's
not
like.
Where
is
it,
then
I
don't
look
always
so
now.
I
look
in
the
right
direction.
It's
it's
more
friendly,
but
unfortunately
I
really
have
to
leave.
That's
so
such
a
pity.
B
Yeah,
have
you
joined
the
the
slack
group
foundation,
yeah.
A
B
Yeah,
so
we
go
through
little
bursts
when
this,
when
there's
things
going
on
right
now,
we've
had
a
lot
of
new
people
joining
in
who
are
just
trying
to
find
their
feet
in
the
locked
space.
B
So
I've
been
spending
a
lot
of
time
just
going
over
the
basic
issues,
but
it
would
be
really
good
to
get
some
time
with
you
to
to
dive
in
a
bit
more
deeply
on
to
some
of
the
challenges
you're
starting
to
find.