►
From YouTube: Incubation Engineering MLOps - Update 2023-02-06
Description
Iterating on UX Review, efforts on Incubation foundations, starting closed testing
- Showcase: https://www.youtube.com/watch?v=E8GhtXXm_O0
- Feedback Issue: https://gitlab.com/gitlab-org/gitlab/-/issues/381660
- This Update: https://gitlab.com/gitlab-org/incubation-engineering/mlops/meta/-/issues/67
- All Updates: https://gitlab.com/gitlab-org/incubation-engineering/mlops/meta/-/issues/16
A
Hello,
everyone
and
welcome
to
another
update
for
incubation
engineering
here
at
gitlab.
My
name
is
Eduardo,
and
today,
I'm
gonna
be
talking
a
little
bit
more.
What's
been
done
on
the
past
few
weeks
for
model
experience
tracking,
so
Circle
back.
What
I
want
to
do
is
give
reasons.
Foreign
tracking
is
one
of
these
I'm
implementing
a
backend
for
ammo
flow,
so
that
data
scientists
can
use
their
ml
flow
code
to
record
their
candidates
or
their
runs
into
gitlab
and
make
it
easier
for
them
to
use
external
tracking.
A
So
on
that
right
now,
I
implement
the
the
past
few
weeks.
I
did
Implement
usage
metrics
for
this.
For
this
feature,
so
a
few
metrics
to
just
got
how
much
users
will
start
using
it.
So
number
of
candidates
created
number
of
of
projects
number
of
users
that
created
candidates,
those
kinds
of
stuff,
then
so
these
are
usage,
metrics
and
but
they
just
tells
usage,
not
customer
value.
A
The
value
proposition
of
this
tool
is
a
little
bit
harder
to
measure
with
our
quantity
with
quantitative
metrics,
because
there
are
more
about
time
to
deploy
a
time
to
help
me
to
use
it
and
later
on
interactions
so
likely
for
now
value
aggregation,
metrics
aggregate
value
metrics
will
be
delivered
as
qualitative
when
we
work
with
customers,
Beyond
implementing
usage,
metric
I
added
I'm
working
on
the
ux
review
feedback,
we
had
I
had
a
ux
review
for
this
feature
and
I'm
working
on
a
great
feedback
that
it
was
given
onto
it
beyond
that,
I
would
invest
a
little
bit
of
time
on
what
I'm
calling
incubation
meta.
A
Well,
we
incubation
Engineers.
We
have
a
different
flow
of
work
because
we
are
creating
a
lot
of
stuff
so
and
we
are
starting
to
create
repeat.
We
are
using
repeated
code,
repeated
patterns
and
I
invest
a
little
bit
of
time
into
extracting
a
bit
of
these
patterns
into
components
or
into
usable
functions
so
that
we
can
share
between
between
us.
Additionally,
I
recorded
a
larger
Showcase
of
model
experiment
tracking.
So
it's
a
12
minute
video
that
explains
what
is
model
experiment
tracking.
A
What
are
we
trying
to
achieve
how
we
are
trying
to
achieve,
and
also
how
it
integrates?
What
does
it?
Why
does
it
make
sense
for
it
to
be
part
of
gitlab?
So,
if
you're
interested
on
this
kind
of
stuff,
the
link
will
be
below
for
the
next
two
weeks.
A
What
I
want
to
do
is
get
wrap
up
the
ux
review
and
get
to
close
the
test
rolling
by
closer
test.
A
I
mean
I
want
to
find
two
or
three
customers
that
are
willing
to
explore
this
with
me
and
are
willing
to
implement,
to
use
gitlab
experiment,
tracking,
a
model,
experiment
tracking,
and
my
intention
is
at
least
for
one
month
or
two
months-
give
full
support
or
full
priority
to
these
two
or
three
customers
and
really
iterate
on
their
feedback
before
I
open
this
for
General
Public,
so
get
a
very
one-to-one
relationship
with
customers
and
act
on
their
feedback
really
fast
and
iterate
all
their
feedback
really
fast.
A
So
this
will
be
happening
in
the
next
two
weeks,
I'm
already
talking
with
with
some
customers
to
get
this
rolling
and
that's
what
that's
all
I
had
for
today.
Thanks
for
watching.