VENCESLAS OSEE NGASSAM KATE

Context & Challenges Role & Key Achievements Results & Impact

• Société Générale managed massive volumes of sensitive financial and customer data for credit scoring and decision support.
• Existing scoring pipelines required modernization and industrialization to enhance model performance, data traceability, and compliance with regulatory requirements.
• The primary challenge was to ensure faster, more reliable, and explainable credit decisions within a highly constrained banking environment (security, auditability, data governance).

• Redesigned and industrialized credit scoring pipelines using Azure Databricks and Azure Machine Learning.
• Designed robust batch pipelines for the preparation, transformation, and validation of sensitive financial data.
• Implemented workflow orchestration via Apache Airflow, including monitoring, alerting, and incident management.
• Collaborated closely with Risk, Business, and IT teams to translate functional and regulatory requirements into operational data solutions.
• Contributed to the deployment and monitoring of Machine Learning models (feature engineering, performance monitoring).
• Participated in data documentation, traceability, and governance initiatives essential to the banking sector.

• Improved credit scoring model accuracy by approximately 15%.
• Reduced delays in credit decision-making processes.
• Strengthened the reliability and auditability of data pipelines

Context & Challenges Role & Key Achievements Results & Impact

• The National Institute of Statistics was responsible for processing national census data, involving massive volumes of demographic, socio-economic, and geographic data.
• Data originated from multiple, often heterogeneous sources, presenting significant challenges regarding data quality, reliability, and statistical consistency.
• Insights needed to be clear and actionable for non-technical decision-makers within a high-stakes institutional and public policy context.
• Analyzed, cleaned, and structured massive demographic datasets (over 10 million records) using Python, SQL, and R.
• Performed exploratory data analysis (EDA) and advanced statistics to identify trends, anomalies, and key indicators.
• Developed statistical and predictive models to analyze demographic and territorial evolution.
• Designed and automated decision-support dashboards (Power BI, Tableau) for institutional leaders.
• Implemented spatial analysis (GIS) using QGIS and Python, cross-referencing geographic data with demographic insights (density, distribution, territorial dynamics).
• Communicated and simplified complex findings for non-technical stakeholders (statisticians, public officials, and decision-makers).
• Improved the reliability and readability of the demographic indicators produced.
• Accelerated the production of statistical reports and decision-support analyses.
• Enhanced the integration of geographic data into national-level analyses

Context & Challenges Role & Key Achievements Results & Impact

• Orange managed large volumes of heterogeneous network data (telecom equipment, logs, real-time events) used for quality of service monitoring and incident management.
• Existing processes were insufficiently industrialized, resulting in high processing times and limited reactivity from network teams when facing critical incidents.
• The challenge was to implement a reliable and high-performance Big Data platform capable of processing massive real-time and batch streams while improving operational visibility.
• Joined the Data team responsible for the supervision and performance analysis of telecommunications networks.
• Designed, developed, and maintained Big Data ETL pipelines for the collection, transformation, and analysis of data from network equipment.
• Processed millions of real-time network events using Apache Spark and the Hadoop ecosystem.
• Implemented batch and streaming processes for incident monitoring, anomaly detection, and Quality of Service (QoS) analysis.
• Optimized pipeline performance, leading to reduced processing times, improved resource management, and increased reliability.
• Collaborated closely with network and IT teams to translate operational needs into actionable data solutions.
• Authored technical documentation for pipelines and data flows to facilitate maintenance and future solution evolution.
• Reduced processing time for critical tasks by approximately 30%.
• Significantly improved the reactivity of network teams regarding incidents and anomalies.