Data Consulting Firm specializing in various sectors, including energy, healthcare, industry, and services.Key Responsibilities:---------------------- ---> Conducted data analysis and modeling using Python and R. ---> Managed data throughout its lifecycle. ---> Developed Computer Vision models. ---> Engaged in Research and Development projects in data science. ---> Created packages and web applications using Python, R Shiny, JavaScript, and Vue.js. ---> Implemented MLOps and DevOps practices, including containerization with Docker, Continuous Integration and Continuous Deployment (CI/CD), automated testing, performance monitoring, and application support and maintenance.

As a Teaching Assistant for practical and tutorial sessions in Machine Learning, I supported students in understanding core concepts and applications of machine learning algorithms. My responsibilities included: ---> Assisting in the preparation of course materials and assignments to reinforce theoretical knowledge through practical application. ---> Guiding students through hands-on exercises and projects, fostering a practical understanding of machine learning algorithms and techniques. ---> Addressing students' questions and helping them troubleshoot issues in their pratical labs.This role allows me to deepen my own knowledge while mentoring the next generation of data scientists and machine learning practitioners.

Development of a Data Orchestration, Qualification, and Validation ApplicationContext:-------- ---> Design and develop an application to orchestrate data flows, ensuring the qualification and reliability of measured data. ---> Collaborate with cross-functional teams to optimize data integration and processing workflows. ---> Provide ongoing application support and maintenance to ensure continuous system reliability and performance.

Consultant for the Financial Directorate - Development of a Web Application Suite for Data Flow Monitoring and Energy Forecast Quality ControlKey Taks:---------- ---> Redesign and rewrite an existing R Shiny application to create a comprehensive ecosystem of R Shiny applications. ---> Develop R Shiny modules for monitoring data flows, tracking model performance, and comparing model outputs. ---> Implement Dockerization and set up a continuous deployment pipeline tailored to the infrastructure (Kubernetes Cluster).

Consultant for the R&D Department - Development of an Algorithm for Medium/Long-Term Forecasting of Electrical Load TrendsKeys Tasks:------------ ---> Conducted an in-depth analysis and diagnostic of the previous forecasting method. ---> Cleaned and analyzed time series data of electric load from source substations. ---> Proposed and implemented a trend forecasting algorithm using quantile autoregression to forecast electric load trends.This work offered robust results and better insights for future planning and investment.

As a data scientist in an R&D team, I focus on validating Generative Adversarial Network (GAN) discriminators for mechanical design by quantifying uncertainty in deep neural network predictions. My work employs Bayesian inference methods in deep learning, primarily using images as data.Key Responsibilities:---------------------- ---> Developed deep learning models for the design of 3D-printable mechanical parts, including image generation and classification. ---> Built training and evaluation pipelines using KALE to streamline model development. ---> Conducted a state-of-the-art review on uncertainty quantification in deep learning and applied the insights to enhance model performance. ---> Implemented advanced techniques for uncertainty quantification, including Monte Carlo Dropout, Bayes By Backprop, Evidential Neural Networks, Stochastic Gradient Hamiltonian Monte Carlo, and Stochastic Gradient Langevin Dynamics.

Semantic Address Matching and AnalysisKey tasks:----------- ---> Automated the extraction of information from postal addresses using Deep Learning (NLP, NER) and Hidden Markov Models. ---> Developed an address matching algorithm with the BD TOPO database using a vector-based database with FAISS. ---> Built a web application with Flask to deploy the constructed model and matching algorithm.

Proposed an effective process for analyzing the impact of treatments.Key tasks:----------- ---> Identified the limitations of using p-values in statistical analysis. ---> Analyzed the prevalence of the hepatitis E virus in pig farms at slaughter age. ---> Implemented bayesian techniques to assess the practical significance of comparison tests.

Redesigned the retail credit score for leasing.Key tasks:----------- ---> Selected and discretized of variables using decision trees. ---> Built scoring models using logistic regression and XGBoost with SAS and R. ---> Developed a scoring grid to evaluate and classify risks more effectively. ---> Improved the bank's scoring model's discriminating power by 10%.

This is an academic internship at the Department of Statistics and Economic Studies of the National Institute of Statistics and Economic Analysis of Benin. It was carried out as part of my bachelor's degree in Statistics and allowed me to produce my final dissertation on the topic: Gross Domestic Product and Electrical Energy in Benin: a causality analysis. It allowed me to deepen my knowledge of time series econometric methods (unit root tests, ARDL, VAR, Toda-Yamamoto causality test).