My Projects
Home
Free AI Personal Statement Generator
Artificial Intelligence Personal Statement
Elena
Vasquez
M
.
S
.
in
Artificial
Intelligence
and
Machine
Learning
My
fascination
with
artificial
intelligence
began
not
in
a
lecture
hall
,
but
in
a
hospital
waiting
room
.
In
2019,
while
accompanying
my
grandmother
through
oncology
appointments
,
I
watched
physicians
struggle
to
reconcile
her
complex
medication
history
across
fragmented
electronic
records
.
That
afternoon
,
I
realized
that
intelligence
,
at
its
most
valuable
,
is
not
about
replacing
human
judgment
but
about
amplifying
it
under
uncertainty
.
I
returned
to
campus
and
changed
my
major
from
biology
to
computer
science
the
following
week
.
Since
then
,
I
have
pursued
machine
learning
as
a
tool
for
solving
problems
that
resist
pure
logic
:
messy
,
high
-
stakes
,
deeply
human
problems
.
As
an
undergraduate
at
the
University
of
Michigan
,
I
built
a
foundation
in
probabilistic
modeling
,
linear
algebra
,
and
distributed
systems
.
In
Professor
David
Chen
'
s
seminar
on
Bayesian
inference
,
I
became
captivated
by
how
prior
beliefs
and
observed
evidence
merge
into
actionable
predictions
.
I
extended
this
interest
through
a
semester
-
long
project
modeling
patient
readmission
risk
for
a
regional
health
system
,
using
Gaussian
process
regression
to
quantify
uncertainty
in
sparse
longitudinal
data
.
The
model
did
not
merely
predict
;
it
flagged
cases
where
the
data
itself
was
too
thin
to
support
a
decision
,
prompting
human
review
.
This
experience
taught
me
that
the
best
AI
systems
are
those
that
know
the
limits
of
their
own
knowledge
.
After
graduating
in
2022,
I
joined
the
NLP
research
group
at
Cohere
Health
as
a
junior
machine
learning
engineer
.
Over
eighteen
months
,
I
contributed
to
the
development
of
clinical
summarization
models
that
extracted
structured
medication
histories
from
unstructured
discharge
notes
.
Working
with
a
dataset
of
2.3
million
records
,
I
implemented
a
pipeline
using
transformer
-
based
architectures
and
weak
supervision
to
reduce
annotation
costs
by
34
percent
.
More
importantly
,
I
led
an
internal
audit
that
identified
demographic
performance
gaps
in
our
extractive
models
,
particularly
for
patients
with
limited
English
proficiency
.
We
addressed
this
by
augmenting
training
data
with
back
-
translated
clinical
text
,
which
improved
F
1
scores
for
that
subgroup
from
0.71
to
0.89.
This
work
solidified
my
belief
that
technical
rigor
and
ethical
vigilance
must
advance
together
.
The
turning
point
in
my
trajectory
came
during
a
collaboration
with
the
hospital
'
s
palliative
care
unit
.
Clinicians
needed
a
model
that
could
predict
patient
mortality
risk
to
inform
early
hospice
referrals
,
yet
they
were
rightly
wary
of
black
-
box
predictions
in
such
sensitive
decisions
.
I
worked
with
social
workers
to
design
a
prototype
that
paired
risk
scores
with
local
,
interpretable
feature
attributions
.
Watching
a
physician
use
our
tool
to
explain
a
recommendation
to
a
family
member
,
I
recognized
that
interpretability
is
not
an
afterthought
;
it
is
the
interface
through
which
AI
earns
trust
.
Since
that
project
,
I
have
oriented
my
research
toward
robust
,
interpretable
methods
that
preserve
the
agency
of
end
users
.
I
am
applying
to
the
M
.
S
.
in
Artificial
Intelligence
and
Machine
Learning
program
to
deepen
my
theoretical
understanding
of
causal
inference
and
robust
optimization
under
distribution
shift
.
I
am
particularly
eager
to
study
with
faculty
working
on
uncertainty
quantification
in
high
-
dimensional
settings
,
as
well
as
the
societal
impacts
of
automated
decision
systems
.
My
goal
is
to
develop
methods
that
make
reliable
,
transparent
predictions
in
domains
where
failure
carries
significant
human
cost
:
clinical
medicine
,
public
health
policy
,
and
criminal
justice
risk
assessment
.
1 / 2
My
immediate
objective
after
graduate
school
is
to
join
a
research
lab
or
mission
-
driven
startup
that
deploys
machine
learning
in
health
care
delivery
.
Within
ten
years
,
I
hope
to
lead
an
interdisciplinary
team
of
engineers
,
clinicians
,
and
ethicists
building
predictive
systems
that
reduce
inequity
rather
than
encode
it
.
The
technical
challenges
are
substantial
,
but
I
am
drawn
to
them
precisely
because
they
are
inseparable
from
questions
of
values
,
governance
,
and
design
.
Artificial
intelligence
is
often
described
as
a
pursuit
of
superhuman
capability
.
My
own
pursuit
is
more
modest
and
,
I
believe
,
more
urgent
:
I
want
to
build
systems
that
know
when
they
do
not
know
enough
,
that
explain
themselves
plainly
,
and
that
leave
the
final
decision
where
it
belongs
,
with
the
people
whose
lives
they
affect
.
This
program
represents
the
ideal
environment
in
which
to
acquire
the
rigorous
training
and
critical
perspective
necessary
for
that
work
.
I
am
ready
to
contribute
fully
and
to
learn
continuously
.
2 / 2
More Examples
