███╗ ███╗ █████╗ ██████╗ ██╗ ██╗ █████╗ ███╗ ██╗███████╗ ██████╗ ██████╗ ██╗ ██╗██████╗ ██████╗ ██╗███╗ ███╗
████╗ ████║██╔══██╗██╔══██╗██║ ██║██╔══██╗████╗ ██║██╔════╝ ██╔══██╗██╔═══██╗██║ ██║██╔══██╗██╔══██╗██║████╗ ████║
██╔████╔██║███████║██████╔╝██║ █╗ ██║███████║██╔██╗ ██║█████╗ ██████╔╝██║ ██║██║ ██║██████╔╝██║ ██║██║██╔████╔██║
██║╚██╔╝██║██╔══██║██╔══██╗██║███╗██║██╔══██║██║╚██╗██║██╔══╝ ██╔══██╗██║ ██║██║ ██║██╔══██╗██║ ██║██║██║╚██╔╝██║
██║ ╚═╝ ██║██║ ██║██║ ██║╚███╔███╔╝██║ ██║██║ ╚████║███████╗ ██████╔╝╚██████╔╝╚██████╔╝██║ ██║██████╔╝██║██║ ╚═╝ ██║
╚═╝ ╚═╝╚═╝ ╚═╝╚═╝ ╚═╝ ╚══╝╚══╝ ╚═╝ ╚═╝╚═╝ ╚═══╝╚══════╝ ╚═════╝ ╚═════╝ ╚═════╝ ╚═╝ ╚═╝╚═════╝ ╚═╝╚═╝ ╚═╝
_, _, _, _,
.' ( .-'./ .' ( .-'./
_/..._'. .'.-'/ _/..._'. .'.-'/
.-'` ` '-./.'_.' .-'` ` '-./.'_.'
( o) )) ;= <_ ~~ Welcome to my terminal-like website! ~~ ( o) )) ;= <_
'-.,\\__ __.-;`\'. '. '-.,\\__ __.-;`\'. '.
\) |`\ \) '.'-.\ \) |`\ \) '.'-.\
\_/ '-._\ \_/ '-._\
Wanna use my graphical interface website? Type [[;rgba(245, 40, 145, 0.99);]"gui"] and press enter.
----
Type [[;rgba(245, 40, 145, 0.99);;]"help"] to see available commands. I've hidden some [[b;rgba(250, 251, 180, 1);;]easter eggs] so feel free to try new commands ;)
Hi ! My name is Marwane Bourdim.
I am a recent math graduate from Université Paris-Cité and Sorbonne Université and [[b;rgba(250, 251, 180, 1);;]I am currently on the job market for data scientist and quantitative analyst positions].
At the moment, I have had the opportunity to do two very enriching research internships:
The first one at INRIA Paris-Saclay in the SIMBIOTX lab where I applied stochastic models to study disease propagation.
The second one at the European Bioinformatics Institute (EBI) in the Cancer Genomics lab in Cambridge (UK) where I used unsupervised machine learning methods
to guess the proportions of different cell types in heteregeneous samples. Right after my second research internship, I became a teacher for
6 months from february to june 2023 at the prestigious Ecole Alsacienne in Paris.
My goal is to use math and computer science for the greater good, learn more about ourselves, and have fun doing it :)
[[b;rgba(250, 251, 180, 1);;]I am also interested in phd projects in machine Learning and mathematics], especially applied to biology and personalised medecine,
or centered around NLP (a soft spot of mine is the CETI project)!
Apart from math, I'm passionate about philosophy, drawing and teaching.
[[b;rgba(250, 251, 180, 1);;]Master's degree in Mathematics, Statistics and Machine Learning]
Université Paris-Cité, France | 2021 ~ 2022
[[b;rgba(250, 251, 180, 1);;]Master's degree in Mathematical Modeling]
Sorbonne Université, France | 2020 ~ 2021
[[b;rgba(250, 251, 180, 1);;]Bachelor's degree in Pure Mathematics]
Université Paris-Cité, France | 2018 ~ 2019
[[b;rgba(250, 251, 180, 1);;]Classe préparatoire aux grandes écoles MPSI-MP]
Lycée Jacques Decour, France | 2016 ~ 2018
[[b;rgba(250, 251, 180, 1);;]IELTS 8]
British Council | 2022
[[b;rgba(250, 251, 180, 1);;]Cellular deconvolution algorithms for nanopore bulk methylation data]
[[i;rgba(168, 218, 141, 1);;]Supervised by Isidro Cortés-Ciriano at European Bioinformatics Institute in the Cancer Genomics lab in Cambridge, United Kingdom | February 2022 ~ January 2023]
[[i;rgba(168, 218, 141, 1);;]Programming language: Python]
[[;rgba(250, 251, 180, 1);;]Semi-technical summary:] DNA methylation is a process by which a certain molecule binds to parts of our DNA.
It plays an important role in the expression of our DNA and thus, different cell types have got different patterns of methylation.
Most of the time, in biology and medecine when we collect a sample, it is made of plenty of different cell types.
When we know the "typical" methylation pattern of given cell types we can lever the fact that different cell types have different methylation patterns
to estimate the proportions of the different cell types in a heteregeneous sample.
During this internship, I developped two different machine learning algorithms to do so, and one can even estimate the proportion of "unknown" cell types.
(those for which we don't have their typical methylation patterns. )
[[;rgba(250, 251, 180, 1);;]Technical summary:] I developped two methylation deconvolution algorithms for Nanopore data.
One supervised maximum-likelihood-estimation binomial regression algorithm
and an unsupervised modified version of NMF to co-estimate the proportions of the known celltypes, the proportion of unknowns
cell types and the average methylation profile of the latter.
[[b;rgba(250, 251, 180, 1);;]Mathematical and computational modeling of the Covid-19 pandemic in France with a spatio-temporal stochastic framework]
[[i;rgba(168, 218, 141, 1);;]Supervised by Dirk Drasdo and Jules Dichamp at INRIA in the SIMBIOTX lab in Paris, France | April 2021 ~ September 2022]
[[i;rgba(168, 218, 141, 1);;]Programming language: Python]
[[;rgba(250, 251, 180, 1);;]Semi-technical summary:] We modeled the different French regions by nodes on a graph and we consider every one of those nodes as a compartment.
Inside of a compartment there are healthy, infected and recovered people. The frequency at which a healthy individual gets infected in a
compartment depends on the number of people in those 3 populations in the compartment. We coded a simulation of this model to study the
evolution of those populations and then added complexity to the model by adding different strands of the virus, a "return-to-home" model for more
realistic individual movements and looked at different type of graph structures to study the influence of the topology on the pandemic dynamic (centralized vs federal country).
[[;rgba(250, 251, 180, 1);;]Technical summary:] We developped a variety of SIR compartment agent-based models via a set of Masters equations and we used the Gillespie algorithm for the simulations.