Senior Data Scientist

• Lead a team of 5 data scientists and a data science community at Raja Group France, the parent company.
• Worked on projects including attribution modeling, Email prospecting, customer segmentation, sentiment analysis, and stock prediction.
• Collaborated with Google on a customer lifetime value project and worked with SAS to set up their cloud solutions, CI360 and Viya.
• Utilized web analytics tools such as Google Analytics, Visual IQ, and IBM Coremetrics, along with Google Cloud Platform products like Big Query, GA3, GA4, and Google Cloud Storage.
• Implemented REST APIs, built SOAP APIs, and used Tableau, SAS Visual Analytics, and PowerBI for reporting.
• Developed a Data-Driven Attribution model from scratch for EU-7 countries to optimize marketing investment across online and offline channels. The custom solution outperformed Google Attribution and now guides all marketing investment decisions.
• Developed a customized algorithm to predict customer lifetime value (LTV), enhancing Google's bidding algorithm for the acquisition of high-value B2B customers.
• Implemented an aspect-based sentiment analysis solution to analyze customer feedback and address high churn rates.
• Developed a model to identify high-potential customers early in their journey based on online and offline interaction data.
• Built a customer segmentation model from scratch based on predictive sales, which is used for all marketing campaigns across 7 European countries.
• Created a stock prediction project for warehouse stock optimization.
• Developed an in-house bespoke product recommendation solution for Viking, rolling out user-based collaborative filtering and currently refining it into a hybrid solution.
• Tools Used: Azure Cloud, GCP, Python, PySpark, SAS, Machine Learning, Deep Learning (Keras, TensorFlow), NLP, Sklearn, PowerBi, Big Query, W2V, GloVe.

As part of my Master Thesis, I used Deep Learning and traditional machine learning approaches to solve a time-series forecasting problem. Specifically, I predicted young soccer players' peak potential with optimal age. I worked mostly with Python using Keras (with TensorFlow backend), Scikit-learn, and other libraries during my thesis.