Last update: 2022-07-18
Hello!
My name is Damien Desfontaines — also known as
Ted, or any pseudonym recognized by the (Ted)+ regular
expression — and this is the sober version of my personal
webpage. The fancier one, which is now quite outdated, can be found
right here.
About me
I'm a senior scientist at Tumult Labs,
a startup focusing on
differential privacy.
Before that, I was leading the anonymization consulting team at Google. My
job is to make it easy to safely anonymize data.
In December 2020, I defended my PhD thesis at
ETH Zürich, which I worked on
part-time in the
Information Security Group, under
the supervision of
David Basin.
Previously, I studied at the
École normale supérieure, where I
completed a Master's degree in mathematical logic and theoretical computer
science called the
LMFI.
Before that, I did the first year of the
MPRI, a theoretical
computer science master.
You can find more information on my
LinkedIn profile.
I'm also @TedOnPrivacy on
Twitter, and
@tedted@hachyderm.io
on Mastodon.
Scientific papers
When I was studying for my Master's degree, I worked on pretty
theoretical stuff.
-
Cristian S. Calude,
Damien Desfontaines.
Universality and Almost Decidability.
Fundamenta Informaticae,
2015
(PDF,
tl;dr: maybe it's possible to approximate the
halting problem!)
-
Cristian S. Calude,
Damien Desfontaines.
Anytime Algorithms for Non-Ending Computations.
IJFCS,
2015
(PDF,
longer version of the previous paper)
-
Laurent Bienvenu,
Damien Desfontaines,
Alexander Shen.
What Percentage of Programs Halt?.
ICALP,
2015
(PDF,
tl;dr: actually, no, it's impossible.)
-
Laurent Bienvenu,
Damien Desfontaines,
Alexander Shen.
Generic algorithms for halting problem and optimal machines revisited.
LMCS,
2015
(PDF,
longer version of the previous paper)
Since then I've been working on
data anonymization.
-
Damien Desfontaines,
Andreas Lochbihler,
David Basin.
Cardinality Estimators do not Preserve Privacy.
PETS,
2019
(PDF,
blog post,
recording of my talk,
tl;dr: when counting people, in order to remember who you've already
counted, you have to remember who you've already counted.)
✨ Best Student Paper Award ✨
-
Pern Hui Chia,
Damien Desfontaines,
Milinda Perera,
Daniel Simmons-Marengo,
Chao Li,
Wei-Yen Day,
Qiushi Wang,
Miguel Guevara.
KHyperLogLog: Estimating Reidentifiability and Joinability of Large Data at Scale.
IEEE S&P,
2019
(PDF,
recording of Pern's talk,
tl;dr: a clever way of estimating how easy it is to reidentify people
in very large datasets.)
-
Royce J Wilson,
Celia Yuxin Zhang,
William Lam,
Damien Desfontaines,
Daniel Simmons-Marengo,
Bryant Gipson.
Differentially Private SQL with Bounded User Contribution.
PETS,
2020
(PDF,
recording of Royce's talk,
tl;dr: a system to anonymize data easily and securely.)
-
Damien Desfontaines,
Balázs Pejó.
SoK: Differential Privacies.
PETS,
2020
(PDF,
recording of my talk,
tl;dr: a categorized list of all variants and extensions of
differential privacy,
a central concept in anonymization.)
-
Gregory A Wellenius,
Swapnil Vispute,
Valeria Espinosa,
Alex Fabrikant,
Thomas C Tsai,
Jonathan Hennessy,
Brian Williams,
Krishna Gadepalli,
Adam Boulanger,
Adam Pearce,
Chaitanya Kamath,
Arran Schlosberg,
Catherine Bendebury,
Charlotte Stanton,
Shailesh Bavadekar,
Christopher Pluntke,
Damien Desfontaines,
Benjamin Jacobson,
Zan Armstrong,
Bryant Gipson,
Royce J Wilson,
Andrew Widdowson,
Katherine Chou,
Andrew Oplinger,
Tomer Shekel,
Ashish K Jha,
Evgeniy Gabrilovich.
Impacts of US State-Level Social Distancing Policies on Population
Mobility and COVID-19 Case Growth During the First Wave of the
Pandemic.
Nature Communications,
2021
(HTML,
tl;dr: what the title says, based on anonmyized data from Google
Location History users.)
-
Damien Desfontaines,
James Voss,
Bryant Gipson,
Chinmoy Mandayam.
Differentially private partition selection.
PETS,
2022
(PDF,
poster
for TPDP 2020,
tl;dr: an optimized technique to publish differentially private
statistics on an unknown domain.)
-
Damien Desfontaines,
Esfandiar Mohammadi,
David Basin,
Elisabeth Krahmer.
Differential privacy with partial knowledge.
Under review
(PDF,
blog post,
tl;dr: what happens to differential privacy when modeling an attacker
with only partial knowledge over the data.)
Other publications
I have a blog, where I mostly write about
anonymization in simple terms. I recommend starting with my
friendly introduction to differential privacy,
for which I've gotten good feedback from a number of folks.
I published my PhD thesis online as a series of
Web pages. Its index highlights the sections that don't have too much math
in them, in an effort to make parts of it accessible for people who aren't
experts.
During the COVID-19 crisis, I helped several efforts to publish
anonymized data from Google users to help researchers and public health
authorities understand and combat the pandemic. For three of these
projects, we published a technical description of the anonymization
process:
In 2020, I was interviewed by the ACM
with my colleague
Miguel Guevara
to discuss our efforts around differential privacy. The interview
(PDF)
was published in
ACM Queue and
Communications of the ACM.
Finally, I write code from time to time. In my previous job, I
contributed to Google's differential privacy libraries, parts of which are
open-source.
More recently, I've been working on
Tumult Analytics,
an open-source library for differential privacy built by my colleagues and
I at Tumult Labs.
Things I like, things I do
- I love to cook! Here are some
marvelous recipes I tried out
for you (in French).
- I like tabletop role-playing games, and I sometimes play or DM
on Dungeons & Dragons.
- When I have some free time, I sometimes write alexandrines. If
you like poetry and speak French, you may have fun reading our version
of the
Declaration of the Rights of Man and the Citizen,
or the abstract of my PhD thesis
(in French).
- I used to work as an oral examiner in mathematics, in the French
high school Louis-le-Grand. I love complicated math problems, especially
when they can be solved using only elementary results. Here are
some of my favourites (in French).
- I love simple programming languages that make it easy not to shoot
yourself in the foot, for example Go
or OCaml.
- I listen to music all the time. I like
trip-hop,
electro,
symphonic metal,
jazz,
classical music,
ska,
hip-hop,
mixes of everything,
sometimes all at once
and it is really hard for me to stop this list without adding
dozens of other links to wonderful artists' websites.
- I like to ski and to admire
snowy mountains
when winter is coming.
To get in touch with me, you can send an e-mail to
se.niatnofsed@neimad (beware when
copying it). I also use WhatsApp and Signal, feel free to ask me for my
number if that's your favorite communication method.
I also love to send and receive postcards. If that's also your thing,
don't hesitate to ask me for my mailing address ♥
