Freelancer profile translated to English.
Back to original languageAbout Boubacar
I am a Data Engineer and very passionate about Data in general (collection, processing, analysis, modeling, packaging, production deployment, etc.). I have carried out missions with high potential in Machine Learning and Data Engineering. I am very comfortable with DevOps and MLOps best practices and the construction of automated end-to-end pipelines (particularly through Git and on the Cloud). I am ready to put my expertise at the service of your projects.
French
Native or bilingual
English
Fluent
Remote only
Primarily works remotely
Experience
- CarrefourData & ML EngineerRETAIL (LARGE RETAILERS)October 2023 - Today (2 years and 9 months)Within the Data team, assisting in the creation and maintenance of new automated processes to feed the category managers' work tool, aiming to optimize national and European assortment (France, Italy, Belgium, and Spain).• Optimization of the assortment algorithm• Investigation of the schemas and structures of different data tables on Big Query to establish logical relationships between them, contributing to better organization and understanding of the project data• Automation of ELT transformation processes with Data Build Tool (DBT) to greatly improve the data flow process and reproducibility of results• Use of Cloud Composer to orchestrate Directed Acyclic Graphs (DAGs) via Airflow• Continuous improvement of the code coverage rate with unit and integration tests (pytest, coverage, and SonarQube) thus ensuring the reliability and robustness of the developed solutions• Factorization, simplification, and improvement of the assortment optimization module corresponding to the DATA backend (Python environment with FastAPI, OR-Tools, etc.)• Automated backtesting of revenue (CA) estimation models during the staging and de-staging of references in stores (taking into account product carry-over and cannibalization)• Analysis and reporting on the distribution of customer needs units (via Looker Studio and Vertex AI)• Agile work with members of the front, QA, etc. teams via the JIRA board• Documentation of all work carried out on the Confluence platform
- CapfiData Scientist | MLOpsBANKING AND INSURANCENovember 2022 - Today (3 years and 8 months)Paris, FranceContext: Within Capfi's VITADATA unit, redesign and creation of new decision support tools commercialized by the company.Achievements:Project 1:○ Development and deployment of a real-time time series forecasting API to better predict phenomena with pronounced trends and seasonality (optimization under time constraints with Optuna)○ In-depth analysis of State of the Art (SOTA) tools in the field of time series forecasting on Python (comparison of models such as Prophet, ARIMA, etc.)○ Construction of the API based on the FastAPI library and containerization of the application via Docker○ Implementation of automatic testing pipelines (unit and non-regression) and deployment of the built containers on the Container Registry service (GCP)○ Implementation of the IAC (Infrastructure As Code) component through Pulumi for automatic deployment of the service on Cloud Run with very precise parameters (maximum number of containers for scalability, maximum number of requests per container, custom domain name mapping, etc.)Project 2:○ Creation of a Flask Dashboard for managing and forecasting used car prices○ Real-time web scraping of information from the "La Centrale" website with BeautifulSoup○ Modeling car prices based on their characteristics (make, model, mileage, etc.) with scikit-learn (use of transformers).○ Tracking ML experiments with MLflow○ Use of Terraform with Gitlab for testing, building, and deploying on GCR and Cloud Run (GCP services)○ Use of SQLite3 and BigQuery for storing articlesProject 3:○ Automatic generation of online educational content for learning Python best practices○ Prompt Engineering and use of the OpenAI API via Python to generate a new lesson to learn from ChatGPT.○ Extraction of code snippets and automatic image generation of the snippet through the Carbon API and Selenium WebDriver○ Automatic publication on a dedicated Twitter page with Tweepy (via Twitter API)Technical Environment: Python, Pandas, Plotly, Scikit-learn, Optuna, Prophet, GCP, Git (Gitlab), IAC, Pulumi, Terraform, FastAPI, Docker, MLflow, BeautifulSoup, Selenium, Flask, SQL, SQLite3, HTML, CSS, BigQuery, ChatGPT API, Twitter API (Tweepy), Ubuntu 22.04, Shell
- SOCIETE GENERALEData Scientist | Data EngineerBANKING AND INSURANCENovember 2020 - November 2022 (2 years)Mission 1:Context: Within the ITIM/DSR team (Société Générale), design, implementation, and production deployment of a failure detection model for a park of application servers. Achievements:• Improvement of IT park maintenance• Reduction of human intervention• Reduction of failure resolution time• Exploration of logs from the concerned applications (logs sent to HDFS via Kafka)• Multivariate, temporal, etc. statistical analyses of logs• Daily use of PySpark for querying, writing, and exploring data (UDFs, broadcast, etc.)• Implementation of the z-score algorithm via PySpark and exploitation in Spark Streaming (Kafka) to alert potential failures in real-time (code optimization under high time constraints)• Generation of automatic emails to alert the right people in case of detected failure• Deployment of the solution in cluster mode for about ten applicationsTechnical Environment: Python, Spark (Batch, Streaming, UDFs), PySpark, Kafka, Pandas, Numpy, Scikit-Learn, Plotly, Linux,--Mission 2:Context: Within the MoSAIC team (More Security with Artificial Intelligence) at Société Générale, assistance in the analysis, design (R&D), and implementation of Machine Learning algorithms for fraud prevention.Achievements:• Implementation of a scoring model for the Security Pass enrollment for preventive detection of classic and instant transfers of a fraudulent nature (scoring under time constraints).• Improvement of ML models across several areas (Paylib, Instant Transfers, etc.)• Statistical analysis and adaptation of different models to new fraud trends (Phishing, Social Engineering, etc.)● Multivariate, temporal, etc. statistical analyses of legitimate or fraudulent transactions● Creation of customer profiles based on browsing and transaction history● Querying, processing, feature engineering of data to develop high-performance scoring models via H2O (Logistic Regression, Lasso, GBM, XGBoost, Random Forest, etc.)● Packaging of models in several formats (H2O, zip, mojo, etc.) reusable by the scoring API.● Monitoring and automation of the display of the status of operation scoring applications through certain KPIs (crontab, HTML, CSS, PySpark, etc.)● Interactions with the business to better understand business issues and fraud scenarios, then analysis and creation of variables adapted to these needs● Formalization of the need to implement new variables in the Big Data ecosystem.● Continuous documentation of tested methods (Git, JIRA, Powerpoint, Excel)Technical Environment: Python, PySpark, H2O, Scikit-learn HBase, Hive, SQL, Pandas, Plotly, Linux, HTML, CSS, Git (Github), JIRA
Recommendations
Be the first to recommend Boubacar
Help this freelancer shine by sharing your experience working together.
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- Data Science EngineerNational School of Statistics and Information Analysis (ENSAI),2020Ingénieur en Data Science
- Master 2 in Computer Science/BiologyUniversity of Rennes 1,2020Master 2 en informatique/biologie