►
From YouTube: IncEng MLOps Update - 2022-07-28
Description
MLOps JTBD and Started work on MLFlow
MLOps JTBD: https://gitlab.com/gitlab-org/incubation-engineering/mlops/meta/-/issues/58
gitlab-mlflow: https://gitlab.com/gitlab-org/incubation-engineering/mlops/gitlab-mlflow
This update: https://gitlab.com/gitlab-org/incubation-engineering/mlops/meta/-/issues/59
All Updates: https://gitlab.com/gitlab-org/incubation-engineering/mlops/meta/-/issues/16
A
Hello,
everyone
and
welcome
to
another
update
for
incubation
engineering.
Mlaps
today
is
july
28th
and
a
reminder
of
the
vision
we
are
trying
to
make
gitlab
a
two
data
scientists
and
machine
learning
engineers
love
to
use,
and
we
started
that
by
working
on
improving
code
review
experience
for
jupyter
notebooks,
and
this
is
the
usage
data
that
we
have
for
it.
A
We
know
we
see
that
in
july
we
are
almost
at
the
end
of
it,
so
we
will
likely
not
reach
the
the
peak
that
we
saw
in
june,
but
that
is
expected,
because
the
number
of
comments
left
in
cool
reviews
in
general
in
july
was
a
lot
smaller
than
the
number
of
cold
reveals
left
in
june
summer.
Vacation
things
like
that.
We're
still
going
above
this
line
over
here,
which
is
a
very
substantial
growth,
it's
per
especially
compared
to
july
last
year.
A
So
that's
still,
I
don't
know
twice
as
much
moving
on
we
worked
on.
We
took
a
step
back
and
tried
to
look
at
envelopes
through
the
job
distributed
lens,
so
we
tried
to
define
what
are
the
objectives
our
users
are
trying
to
accomplish
by
look
when
looking
at
envelopes,
so
the
basic,
the
most
basic
job
should
be
done.
The
one
mlops
itself
job
to
be
done
is
when
working
with
software.
That
includes
machine
learning.
A
A
From
this
job
to
be
done,
we
can
derive
a
lot
more,
so
this
is
a
short
list
of
36
that
I
was
able
to
come
that
I
was
able
to
come
from
my
experience
and
from
talking
to
customers
spreading
out
through
all
different
stages
of
the
devops
life
cycle.
I'm
trying
to
match
this
to
the
devops
life
cycle,
because
mlabs
is
devops
in
the
end,
so,
for
example,
create
has
it
done
verify?
A
Has
it
done
and
plan
and
secure
only
have
one,
but
that's
mostly
because
my
perception
on
this
areas
is
small,
it's
not
as
complete
as
in
the
other
areas.
This
got
a
lot
of
reach
already
a
lot.
It
generates
very
interesting
conversations
with
the
community.
I'm
very
excited
about
this
and
we'll
be
anchoring
our
explorations
in
onto
this
job
distributor.
A
Moving
on,
we
started
work
on
number
flow.
Yay
we
created
a.
We
are
creating
a
new
component
gitlab
emma
flow,
which
will
be
deployed
along
gitlab.
So
when
you
install
your
gitlab
you're
gonna
install
mlflow
together.
If
you
want
to
do
so,
and
our
mvp
we're
planning
this
eight
to
complete
this,
a
job
should
be
done
over
here.