Anass Bellachehab Email & Phone Number

Issy Les Moulinaux

December 2016-June 2017: Predictive maintenance and anomaly detection in société générale’s IT systemsThe project is part of the big data department’s strategy to explore machine learning solutions forpredictive monitoring of online banking platforms.The project plan has been the following:• Identifying relevant metrics for performance monitoring • Data cleaning and feature identification• Implementing deep learning autoencoder method on historical data.This lead to 100% of application meltdowns being detected 5 minutes in advance and no falsepositives. Our solution has been adopted by the fraud detection team as well for detectingfraudulent transactions.• Presenting the results to project manager using R shiny and python bokeh HMIs.• Deploying the solution to production using Spark• Technical environment: Rstudio, Shiny, H20 ML library, python(Numpy, Pandas,tensorflow), SparkJuly 2017 - August 2018 DATA ENGINEER OrangeProject: Data engineering consultantAs a data engineer, my role is to use big data tools like Hadoop, Hive, Apache Spark, etc. To developproduction ready projects based on preliminary studies done by the data scientists of Orange BIdivision. Some of my tasks include:• Assisting in platform migration from Hadoop 2.4 to Hadoop 2.6• Industrialization and deployment of machine learning solutions using pySpark, Hive and Oozie.• The ML solutions were rewritten in in pySpark using spark data-frames, this led to a 100xspeedup in runtime.• Technical environment: Python, Shell, Hadoop, Hive, Java, Maven, Jenkins, Spark, ScalaJuly 2018​ to June 2019​, July 2018​ to June 2019​ EDFMonitoring and debugging energy consumption forcast algorithms in productionRequirement specification and impact study of GDPR compliance regarding user consent on existing data pipelinesThis contributed to my department being cleared during GDPR compliance auditTechnical environment: Oracle, SQL dev, Hadoop, Hive

Evry

Subject: ”The problem of consensus in distributed sensor networks: the case of nonlinear data”• Proposing a consensus algorithm for sensor networks in the case where stored data does notsupport addition or averaging operations• Estimation the algorithm’s convergence speed• Making simulations of some key examples using python and matlab• The results were published in papers and presented at international conferences• Journal papers:– Entropy: Distributed consensus for metamorphic systems using a gossip algorithm forCAT(0) metric spaces– IEE Transactions on automatic control: Random pairwise gossip on CAT(k) metricspaces• Technical environment: Python(Numpy, Scipy), Matlab

Engineer'S Degree, Electrical And Electronics Engineering

Satellite Communications; Quantum information theory.

Master'S Degree, Physics

Fundamental concepts in solid state physics

Master'S Degree, Applied Mathematics

Stochastic and statistical modelling.

Doctor Of Philosophy (Ph.D.), Applied Mathematics