Hamza Berrehrah

LPCR (Les Petits Chaperons Rouges) is a pioneer in the reception and development of young children. The LPCR group is today the benchmark player in the private nursery sector in France, with more than 850 of its own nurseries and more than 2,800 partner nurseries. In 2022, LPCR acquired the LIVELI player, of equivalent size, made up of the Crèche de France (CDF) and Crèche Attitude (CAT) brands. Projects to pool the IT department and IT and Data tools were then launched. In this completely disrupted environment, several Data and BI projects have emerged, and for which my contribution has been notable. These include: • Construction of a new Data/BI infrastructure and architecture, capable of supporting the large volume and growing size of the company • Migration of on-premise resources to a hybrid/Cloud ecosystem (Microsoft Azure) • Unification of BI tools and migration of the various tools (QlikView, Jedox, MyReport) to a Power BI tool • Redesign of the Data Warehouse and Datamarts, initially built with Power BI Services (Power Query, Power Automate, SharePoint) to SQL Server, SSI and subsequently to Azure Fabric • Redesign of existing Dashboards under Power BI, Qlik View, Jedox • Flattening of the various existing business indicators and KPIs and their reconstruction in collaboration with the various business partners (sales, operations, CDG, accounting, ADV, ...). • Unification of business tools (Salesforce, ERP, Repository, Accounting, etc.) and the data generated Technical and Functional Environment - Business Solutions: Salesforce, ERP, Sage, Navision, Cegid - BI Applications: Power BI, QlikView, MyReport, Jedox, etc. - Data Environment: On-Premise DWH/DM and DWH/DM on Azure - BI Tools: Power BI, Power BI Services, Power Apps, Power Automate - ETL: Power Query, SQL, SQL Server, SSIS, SharePoint - DBMS: SQLServer, MySQL, Postgre - Cloud: Azure, Azure Fabric - OS: Windows, Linux - Project Management: Teams, Jira, Notion, OnNote, Agile and Lean Methods (Scrum, Kanban, SAFe)

In parallel with the above in-depth projects, several project management and other operational tasks were carried out. Here is a summary of these tasks carried out: • Implementation and execution of an agile working methodology with the various developers, business project managers, infra, sponsors, etc. • Implementation and execution of a BI/Dashboards project development strategy, developed with the business lines and the IT department • Implementation of Club Data and communication actions with the business lines on Data/BI projects • Agile management of a team of Power BI/QlikView developers • Reporting of the roadmap, monitoring of projects and developments with the IT department and business sponsors • Development of interactive dashboards and Power BI reports to provide a clear and concise visualization of the company's key data • Implementation of data flows to extract, transform and load (ETL) data from different sources, thus ensuring the consistency and accessibility of crucial information. • Training end users to ensure successful adoption of Power BI applications and reports, while providing ongoing technical support to resolve issues and optimize processes. • Developing Power Apps to automate business processes, improving operational efficiency • Enhancing existing solutions by monitoring user feedback, implementing updates, and optimizing application and report performance. • Contributing to optimizing business operations and improving data-driven decision making. Technical and Functional Environment - Business Solutions: Salesforce, ERP, Sage, Navision, Cegid - BI Applications: Power BI, QlikView, MyReport, Jedox, etc. - Data Environment: On-Premise DWH/DM and DWH/DM on Azure - BI Tools: Power BI, Power BI Services, Power Apps, Power Automate - ETL: Power Query, SQL, SQL Server, SSIS, SharePoint - DBMS: SQLServer, MySQL, Postgre - Cloud: Azure, Azure Fabric - OS: Windows, Linux - Project Management: Teams, Jira, Notion, OnNote, Agile and Lean Methods (Scrum, Kanban, SAFe)

Within the ACV (Alliance Connected Vehicles) department, and with a mixed Japanese-French-American team, based on 3 sites (Paris, Tokyo, San-Francisco (Silicon Valley)): • Submit new ML projects and integrate ML pipelines into the global architecture of the alliance • Data Architecture/engineering on AWS. Incorporation of ML into an AWS architecture. • Intensive participation in different projects, at different stages (experiment design, model building, results discussion,…). • Lead major projects (VPA, UBM,…), from data collection to algorithm deployment, involving different groups at Renault, Nissan and the alliance. UBI (Usage Based Insurrance) • Participate in the development of the strategy and the development of the UBI project in Europe • Study the UBI development in the USA. Clear overview on the current solutions of Nissan and Renault • Contribution with insights to the Alliance's commercial strategy / commercial services design groups on UBI. HMI usage analysis, product improvement with data - VPA (Virtual Personal Assistant) • Data Engineering o Data collection and transform data in Renault Datalake (Hive, Spark) o Prepare data for algorithm development o Develop an automatic pipeline to transform Rlink data (in-car application) into usable data - Develop a dashboard for monitoring/visualizing the VPA database • Algorithm development o Modeling and prediction of the driving context. Traffic type and road type classification with only CAN data from the car, o Develop a rule-based algorithm to suggest an optimal driving mode (4 modes), based on the driving contexts o Segmentation algorithm for air conditioning usage categories • Dashboard and application development o Data monitoring dashboard: develop a Dashboard with Tableau to monitor data quality and data sent to the Datalake from the test car fleet o Driving context visualization dashboard o Mex mode selection dashboard: visualize the Mex mode associated with a driving context and driving behavior

Within the ACV (Alliance Connected Vehicles) department, and with a mixed Japanese-French-American team, based on 3 sites (Paris, Tokyo, San-Francisco (Silicon Valley)): Anomaly detection in connected car logs • Study of the overall architecture of the micro-services of the connected car software platform • Generation of log data and analysis of patterns (millions of logs per day) • Parsing and extraction of interesting features in log data • Development of different types of anomaly detection algorithms: HMM (Hidden Markov Model), LSTM, CPT, statistical methods, etc. • Determination of anomaly log scoring models and the “Health Scan” program. The implementation of these programs and others developed by members of our group has resulted in an estimated ROI of approximately €2 million per year for intensive computing and storage expenses in the Azur cloud. Car Digital Twin • Participate to the Car Digital Twin features and functionnalities • Communicate on to the deployment of the PoC of Nissan Car Digital Twin among Renault departments Predictive Maintenance (UBM: Usage Based Maintenance) • Develop a rich and comprehensive scientific literature review and state of the art (research and industry) for predictive maintenance in the automotive field and “UBM”. • Develop different types of models and algorithms (logistic regressions, ML techniques, Bayesian methods, survival analysis, ...), for tire wear, with different levels of accuracy and complications. Driving behavior (with a driving behavior score), driving environment and car conditions are considered as variables correlated to tire wear • Develop algorithms for brake pad wear, with different levels of accuracy and complications (using FVT (Fleet Validation Test) data) • Develop ML pipeline PoC in AWS architecture.

Within the ACMS (Alliance Connected Vehicles and Mobility Services) department, in the “Big Data Analytics for ACMS” team, a mixed Japanese-French team, based on 3 sites (Paris, Tokyo, San Francisco (Silicon Valley)): • Lead data analytics solutions for Renault Alliance, Nissan and Mitsubishi in the field of connected vehicles and mobility services • Set up and structure the Big Data team in Paris, France • Participate in the communication and evangelization of Big Data Analytics and Data Science projects at ACMS and Renault • Collect sufficient data needed to carry out Big Data analytics projects PoC and prototype for ZOE Dashboard • Collect data from 7 different databases • Design of the Dashboard with Tableau. The online version of the Dashboard is on Tableau Server. • Communication with BtoB at Renault on the Dashboard. Presentation to top management. PoC for Renault Mobility (car rental company) • Collect and clean data • Develop algorithms for car rental take-up optimization and forecasting • Present results to Renault Mobility sales team PoC for app feedback analysis • Design business use cases with ACMS services team • Discuss algorithm choices and sentiment analysis and topic modeling evaluation based on app feedback and comments data • Present results to ACMS services team PoC for Renault e-reputation and ‘’Topic Modeling’’ • Collect and analyze data quality and granularity • Develop and evaluate topic modeling algorithms to model tweet content and topics • Present results and discuss their relevance with Renault e-reputation team Technical and Functional Environment - Business areas: connected/autonomous cars, connected mobility, automotive insurance/finance, etc. - Business solutions: Sales Force, ERP, Datalake/DWH/DM (Renault, Nissan), VPA, HMI - AI tools: ML/DL, SageMaker, NLP/NLU, Flask API, Databricks, Dataiku - BI tools: Tableau, Power BI, Spotfire, Data Studio - Language/ETL: R, Python, SQL, Talend - Data environment: On-premise DWH/DM and DWH/DM on AWS/Azure/GCP - Cloud: AWS (S3, E2, SageMaker, etc.), GCP (Storage, BigQuery, etc.) - Monitoring tools: Splunk, Dynatrace, Grafana - Project management: Teams, Slack, Jira, OnNote, Agile and Lean methods (Scrum, Kanban, SAFe)

Project management in Data Science in different fields: Marketing, Retail, Energy, Automotive Matos Project • Technical framework, expertise and expression of needs; Data Engineering; Model development; Model deployment • Project: Automatic processing of regulatory documents in the construction industry - Automatic verification of the type of document and detection of the quality of its image - Automatic classification of the type of document • Automatic extraction of information in a document Sagemcom Project • Creation of PoCs of segmentation algorithms and automatic separation of scanned documents via the implementation of Artificial Intelligence • Integration into the internal platform and into Demat’Box of an automation of classifier and cutting of documents (invoice, justifications, social, etc.) of documents. Enerdigit Project On behalf of RTE (Electricity Transmission Networks) and based on 4 years of data collected on energy consumption (server group), we were asked to create tools for temporal prediction of electricity consumption of industrial sites in 1/2 hour steps each week based on 7 rolling weeks of data. • Determination of the neural network model with the best efficiency • Implementation of deep learning algorithms (LSTM) • Evaluation of the model by K-fold cross-validation and the use of Grid Search • Predictions are greatly improved compared to mathematical tools (extrapolation, projection, etc.) which do not take into account climatic hazards and the specificities of each site. We achieve an efficiency of 90% and 95% for some sites.

Project management in Data Science in different fields: Marketing, Retail, Energy, Automotive FDJ Project Graphic and visual representation of the e-reputation of certain products, through a social listening program. Use of topic modeling algorithms and dynamic visualization of clusters • Development of a Tableau Software Dashboard to visualize the dynamics of clusters • Modeling of topics and creation of scripts to update the visualization under Tableau. SMeRia Project SMeRia is a mobile application and a web platform that makes life easier for consumers in managing their portfolio and receiving promotions based on their profile and request. Based on the customer's transaction history, the server (uses machine learning algorithms) sends personalized promotions to customers. Summary: improvement of SMeRia functions through machine learning: • Categorization of customer databases, with K-means algorithms • Implementation of an OCR for extracting transactions • Image processing and extraction of different information, with Deep Learning algorithms • Implementation of customer profile and promotion matching algorithms • Design of automatic brand segmentation algorithms (CNN) and customer clustering (deep learning SOM) Bankia Project Bankia is an application that automatically categorizes banking transactions. It thus enables a new range of adapted and personalized banking services. • Implementation of Deep Learning algorithms (CNN) • Dynamic data visualization with Tableau Software • Development of learning algorithms and predictive scenarios of customer behavior Technical and Functional Environment - Business Solutions: Various sectors: Retail, Energy, Finance - AI Tools: ML/DL, NLP/NLU, Flask API, OCR - BI Tools: Power BI, Tableau, Data Studio - Language / ETL: R, Python, SQL - Data Environment: On-Premise DWH/DM and DWH/DM on AWS - Cloud: AWS (S3, EC2, SageMaker, etc.) - Project Management: Teams, Slack, Jira, Notion, OnNote, Agile and Lean Methods (Scrum, Kanban, SAFe)

This visit was in continuation with my research topics of my previous post-doctoral fellow at FIAS. I pursued my: ¤ Particle identification by boosted decision trees (AdaBoost) ¤ Model development, optimization and analysis ¤ Development of a numerical code (Fortran, Mathematica) from effective particles interactions ¤ Constraints study and parameters optimization applicable to elementary particles effective interactions ¤ Data analysis and interpretation of experimental results for QGP hard probes (charm and beauty)

I have worked in the PHSD group from FIAS on a very exciting and interdisciplinary field which involves the modeling of particles diffusion in the thermal bath and their stochastic propagation within a dynamical model following microscopic description. Research Highlights: ¤ Build theoretical models for transport phenomena and particle diffusions in the matter under extreme conditions of temperature and pressure, called Quark Gluon Plasma (QGP) ¤ Developed a theoretical approach to characterize the properties of the matter under extreme conditions of temperature/density ¤ Realization of a fully temperature and density-dependent model (based on the Hard Thermal Loop approach) for the propagation of heavy quarks in a plasma of quarks and gluons ¤ Developed a new original model for nuclear phase transitions ¤ Perform transport simulations for heavy ion collisions ¤ Present and defend our theoretical models and experiment analysis strategies in various meetings and international conferences (more than 18 oral presentations and 2 Posters) ¤ Publication of scientific papers and internal notes (more than 10 articles (5 as first author), published in international Journals) Environnement technique et fonctionnel - Solutions métier : Physique fondamentale, Physique nucléaire, Mathématique, Théorie de transport, Modèles théoriques et phénoménologiques - Outils IA : ML/DL (Arbre de décision, NN,...) - Language : Mathematica, MatLab, C/C++, Fortran

Main Achievements: ¤ Co-Developed stochastic models for particles probing the QGP ¤ Performed perturbative calculations of bound state systems diffusion probabilities in thermal bath

My research sought a theoretical and phenomenological framework to study the bound states (Quarkonia) stochastic propagation and collectivity in a thermal bath (Quark Gluon plasma medium). I co-developed an innovative approach which goes from a microscopic description to the macroscopic observable quantification. My research highlights the role of elastic diffusion processes which were never considered in the literature before. Key Achievements: ¤ Co-Developed a theoretical approach (based on Monte-Carlo methods) describing the propagation of quarkonia in an ultra-relativistic heavy-ion collision ¤ Stochastic localization and dynamic description of particles ¤ Mathematical modeling of bound states scattering probabilities using theoretical tools from diffusion theory, nonlinear dynamical system theory, perturbation theory, … ¤ Data analysis with Mathematica/ROOT/GLE and comparison of the model to the experimental results (SPS, RHIC, LHC) ¤ Writing scientific articles and internal notes. Redaction of my PhD report (450 pages) ¤ Presentation of my work in national and international conferences (more than 10 oral presentations and 3 Posters) Orthogonal Skills: ¤ Data analysis using computing clusters (data mining). Extract useful information from experimental data ¤ Theoretical and phenomenological model development, optimization and analysis ¤ Rapid integration in an international team with a complex technical context ¤ Project Management and international collaborations ¤ Programming skills: C, Fortran, Mathlab, Mathematica, ROOT… ¤ Course attendance of 3MPL Doctoral school (Molécules, Matériaux en Pays de la Loire) ¤ Member of Organizing Committee for doctoral school scientific days (JED2010, ~ about 150 PhD student, elaborating the program, registration list, etc) ¤ Participation in the Logistic organization of an International workshop (QM2011, ~ 900 participant) Technical and Functional Environment - Business Solutions: Fundamental Physics, Nuclear Physics, Mathematics, Transport Theory, Theoretical and Phenomenological Models - AI Tools: ML/DL (Decision Tree, NN, etc.) - Languages: Mathematica, MatLab, C/C++, Fortran