Bertrand B.

Data Scientist

690 dollar

My experience

More

SysnavJanuary 2017 - December 2019

Development of an activity recognition algorithm for ankle-mounted and wrist-mounted inertial device recorder with an innovative machine learning approach

• I collected and organized the data set

• I computed relevant features for the task based on topological data analysis, dynamic time warping and functional data analysis

• I computed an efficient prediction function combining gradient boosting and deep learning (CNN, LSTM, dense network) algorithms

• I optimized the code and developped an executable program

Currently used in clinical studies (supported by the OpenHealth Institute):

• Parkinson: tremor and dyskinesia crises detection

• Duchenne myopathy: stairs and running detection

Supervision of a team of 6:

• I attended a formation of management from SLP management

• I was in charge of recruitment: I designed technical tests and I led the interviews

More

SysnavApril 2016 - January 2017

Development of a stride detector algorithm for an ankle-mounted inertial device recorder with an innovative machine learning approach

• I collected and organized the data set

• I computed relevant features for the task based on patterns detection and functional data analysis

• I computed an efficient prediction function combining gradient boosting and deep learning (CNN, LSTM, dense network) algorithms

• I optimized the code and developped an executable program

• I monitored the code maintenance for clients

More

Safety LineSeptember 2014 - July 2015

Development of a supervised learning algorithm applied to the flight data of black boxes I worked with a PhD student in Data Science co-developing the algorithms used for his thesis based on:

• Functional data analysis: wavelet, principal components analysis

• Features selection algorithm: permutation feature importance

• Random Forest

More

Probability, Statistics and Modeling Laboratory (LSPM)May 2014 - August 2014

Development of a Bayesian PAC algorithm for clustering with Gaussian mixtures

I developed the algorithm based on the work of B. Michel and S. Gaiffas ”Sparse Bayesian Unsupervised Learning”:

• Variable selection and estimation of the number of clusters based on Gaussian mixture models in high dimension

• Metropolis-Hastings algorithm with clustering-oriented greedy proposal

My stack

Technologies

Machine Learning

Big Data

Big Data

Others

Spanish, Data Science

Databases

MySQL

Languages

C/C++, Python, Matlab

Machine Learning

Scikit-Learn, Neural networks, Keras, Deep learning, TensorFlow

IT Infrastructure

Git

Other

Research Engineer in Data Science, C Programming Language, English, Variable selection, functional data analysis, topological data analysis, R, Information Processing, Diploma > Diploma Mathematics, code maintenance, Python Programming, SLP management, Data Scientist, French, Masters Degree > Masters Degree Statistics, German, Doctoral Degree > Doctor of Philosophy > Doctor of Philosophy Data Science, Machine Learning Algorithm Developer

My education and trainings

Master’s degree, Statistics – Data Sience & label Big Data, with honours - University Pierre and Marie CURIE2013 - 2016

Engineer’s degree, Mathematics applied to signal and information processing - Supélec2010 - 2014