Rémi Varloot
Rémi
 Varloot

Research

Network and Service Automation

I am a research engineer at Nokia Bell Labs France in Paris-Saclay. My research focuses on network and service automation in home, industrial and corporate networks, namely with regard to the Internet of Things. This includes Software Defined Networking, network monitoring and analytics, and service discovery and automation.

Rapid Mixing of Local Graph Dynamics

Most of my PhD consisted in studying the mixing time of particular families of dynamic graphs. These were characterised by local edge dynamics for which the stationary distribution gives expanders with high probability. Such distributions are commonly used to model actual networks, such as social or peer-to-peer networks.

For a well-chosen dynamic over graphs with n vertices, we have shown that the mixing time is of order O( n logk n ). This implies that there are O( logk n ) updates per vertex, i.e. that the dynamic converges rapidly.

Perfect Sampling

This work consisted in devising a variant of the coupling from the past algorithm. The aim was to reduce the coupling time by skipping events: only active events which alter the state of the bounding chain are considered.

The algorithm is given explicitly, alongside a proof of its correctness and upper bounds on the improved coupling time for sampling from the stationary distribution of some toy queuing models.

Others

Majord'Home Architecture

I worked at the Alcatel-Lucent Bell Labs on the elaboration of the Majord'Home architecture, an SDN approach to delegating the management of home networks to ISPs.

French Internet Resiliency

I worked for a bit for the French Internet Resiliency Observatory at the French Network and Information Security Agency (ANSSI). I implemented an algorithm that constructs a BGP-level map of the French Internet. This was used to study its resiliency and detect critical ASes.

Peer-to-peer super-scalability

I was in charge of the experimental results for a paper on the scalability of peer-to-peer networks.

Network tomography

I used new compression-based methods for clustering in the field of network tomography.

Teaching

Université de Paris

Internet of Things

This class aims to present the different components of IoT solutions, including the specificities of constrained hardware and the architecture of cloud-based IoT solutions.

École Normale Supérieure

Network Models and Algorithms

This class aims at studying the different mathematical and algorithmic techniques for modeling and studying networks of varying nature: communication networks, social networks, energy distribution networks, etc.

Random Structures and Algorithms

This class introduces the basics of probability theory and its applications to certain aspects of computer science: algorithms, communications networks, etc.

Start of Year Projects

First year students are given 2–3 weeks to complete a simple project in groups of 2 to 4. This is generally done under the supervision of a PdD student or a lecturer.

Dancing Links / Algorithm X

The aim was to implement Knuth's Algorithm X using dancing links, and to apply this to problems such as the brute-force solving of Sudoku.

Traffic Modeling

The aim was to simulate traffic in a grid road network, and to empirically determine the arrival threshold at which deadlocks appear.

Supervising Entry Exams

I was one of the students in charge of installing and supervising the practical computer science sessions for the ÉNS entry exam.

I namely wrote up two documents to help students: one describing how to set up a similar environment on their home computer to practice, and a "cheatsheet" of useful commands, which they were allowed to have with them during the exam. Though the exam changes over the years, these should still be somewhat relevant.

About Me

Computer Skills

Programming Languages
HTML5/CSS3/JS
Frameworks

Vue.js AngularJS

Bootstrap Semantic UI

D3.js Three.js reveal.js

Preprocessing

webpack Gulp

TypeScript

Pug Less Sass

NodeJS

Express Socket.IO

Python

Numpy Scipy

Scikit-Learn PyTorch Pandas

Jupyter Notebook/Lab

Other Languages

C C++ C# Java

OCaml Matlab Scilab Maple R

Systems and Software
Virtualization

Docker Virtual Box

Linux

Ubuntu Alpine OpenWRT

iptables OpenVSwitch

Microservices

nginx MongoDB

RabbitMQ MQTT

Other Skills
Office Tools

Microsoft Office, Teams & SharePoint

LaTeX VS Code

Git GitLab GitHub

Spoken Languages

Fluent in French and English

Education and Experience

Nokia Bell Labs France
Since 2018
Enterprise Network Department

Formerly the IoT-Control Department

PhD in Computer Science
2018
PhD Graduation

Supervised by Laurent Massoulié, Marc Lelarge and Ana Bušić

PSL University, Microsoft Research-Inria Joint Center

2015
ÉNS Graduation

 

Pre-Doctoral Experience
2014
Alcatel-Lucent Bell Labs

5 months

2013
ANSSI

6 months

École Normale Supérieure
2013
Master's Degree

Parisian Master in Computer Science (MPRI)

2013
Internship at Dyogene (Inria)

4½ months

2012
Internship at Trec (Inria)

4½ months

2011
Bachelor's Degree

Computer Science Department

2011
Internship at the University of Melbourne

2 months

Previous Education
CPGE (MPSI/MP*)

Lycée Hoche

2008
International Scientific Baccalauréat

Lycée International de Sèvres

International Option: English

Specialty: Mathematics

2008
International Mathematical Olympiades

Honorary Mention

2008
First prize at the Concours Général de Mathématiques

French national exam

Paris Model United Nations

3 years

Harvard Model Congress Europe