►
From YouTube: SEG MLOps Update - December 24th 2021
Description
- User Personas
- Progress on Jupyter Notebook Diffs
All updates: https://gitlab.com/gitlab-org/incubation-engineering/mlops/meta/-/issues/16
A
Hello,
everyone
and
welcome
to
another
session
on
mlaps.
I
am
eduardo,
I'm
the
incubation
engineer
for
mlaps
at
gitlab
and
today's
going
to
be
a
quick
update.
Last
few
weeks
were
a
little
bit
slow
because
I
was
off
and
so
not
a
lot
of
a
lot
of
was
done
since
last
update,
but
two
things
are
worth
mentioning.
A
First
of
all,
one
of
the
reason,
one
of
the
things
that
we
can
do
now
to
help
raising
the
use
case
to
other
teams,
is
the
creation
of
user
personas
for
deputy
for
data
scientists
for
machine
learning
engineer
for
data
engineers.
A
The
thing
is:
I've
been
mentioning
this
before,
but
this
was
never
the
use
case
for
gitlab
or
it
was
not
thought
out
for
this
use
case
and
there's
no
user
research
on
machine
learning
engineers,
there's
no
user
research
on
data
scientists
and
it's
make
really
hard
to
communicate
with
and
to
justify
some
investments
with
other
teams.
Having
a
once,
we
start
doing
this
user
research
for
the
for
these
personas,
then
it's
easier
to
communicate
with
other
teams.
Okay,
why
are
we
doing
jupiter
notebook
devs?
Why
are
we
implementing
these
different
pipeline
strategies?
A
Why
are
we
investing
in
this
or
that
area?
So
I
think
this
is
one
of
the
things
that
we
want
to
work
on
really
hard
next
year,
the
beginning,
because
this
can
pay
off
a
lot
later,
otherwise,
we'll
always
have
it
be
having
to
convince
people
of
other
teams
of
what
we're
doing
here.
A
So
I
created
this
issue.
There
is
a
whole
process
to
create
a
new
persona,
but
this
was
just
a
kickstarter
to
kickstart,
the
pro
the
the
idea
and
within
this
I
believe
there
are
three
main,
very
different
audiences,
the
first
being
the
machine
learning
engineer,
a
software
engineer
that
is
specialized
in
putting
machine
learning
models
into
production.
A
And
then
you
have
the
data
scientist,
which
is
mostly
focused
on
delivering
value
from
data,
either
as
machine
lear
as
with
machine
learning
models
or
with
analytics.
And
then
you
have
to
do
the
engineer,
which
is
a
platform.
Almost
a
platform
engineer
focused
entirely
on
how
to
get
data
from
a
to
b
and
how
to
make
it
easily
accessible
and
the
security
around
the
data,
privacy
and
so
on
and
so
forth.
So
these
are
the
three
persons
that
I
propose
over
here.
A
This
will
take
a
while,
but
it's
a
kickstart
and
the
second
thing
that
I've
been
working
on
it's
the
option
to
toggle
between
roy
and,
if
render
and
raw
diffs.
So
this
has
been
going
on
for
a
while,
and
it
will
still
be
going
on
for
a
couple
for
at
least
a
couple
of
weeks
or
more.
A
So
the
idea
is
that
some
users
do
need
both
the
raw
and
the
users
like
the
render
different
notebooks,
but
some
of
users
also,
but
there
are
some
use
cases
for
the
the
raw
diff
as
well,
because
the
rendered
hide
some
things
that
might
be
useful
for
for
some
users.
So
you
want
to
display
both
of
them,
and
the
challenge
is
that
how
do
I
place
the
comments
in
the
right
place?
A
So
if
I
do
a
comment
on
line
a
on
the
raw,
what
would
be
the
rod,
the
right
line
for
it
to
be
displayed
on
the
transform
this
on
the
render
diff
so
we're
working
on
this?
It's
it's
an
important
change
and
I
think
it's
I
believe
it's
it's
really
really
pay-off,
but
it's
a
very
quiet
progress.
I
would
say
so
it's
a
lot
of
work
on
the
on
the
underlying
library.
A
That
is
not
visible,
so
I
I
have
nothing
to
show
off
a
difference
into
the
y
at
this
point
because
I'm
working
on
the
on
the
on
the
engine
of
the
diff,
but
this
will
come
soon
we're
doing
good
progress
over
here,
we're
almost
at
a
point
where
we
can
already
do
the
the
mapping
between
raw
and
rendered
line
numbers.
So
we
built
the
symbol
map
for
that
and
things
are
progressing
smoothly
over
here.