Walid A.

R&D – Production (End-to-End) | Computable Contract Team Building AI solutions that transform complex insurance contracts into actionable, digital knowledge. By combining knowledge graphs with large language models (LLMs), we create systems that can understand, connect, and analyze contract data at scale. This approach delivers insights that strengthen decision-making across key insurance areas such as underwriting, claims processing, and portfolio analysis.

R&D – Production (End-to-End) | E-commerce Applied Data Science and Operations Research to support growth and enhance operational efficiency in an e-commerce company. Focused on optimizing product category allocation across warehouses to reduce logistics costs and on running A/B testing campaigns to evaluate and improve marketing strategies.

R&D - Institut Roche Master Thesis : Evaluation of anonymization and synthetic data generation methods in clinical studies Working on data privacy issues and (try to) solve them with state-of-the-art generative AI’s algorithms. The goal is to find a synthetic data generation approach that preserves the statistical properties of the original dataset while being perfectly anonymous. This problem is complex because the two criteria are antagonistic, and a compromise must be found between privacy and utility. Key Contributions : • Tested and compared state-of-the-art methods: Gaussian copulas, Bayesian networks, CT-GAN and ARX. • Designed an effective synthetic data generation approach based on information theory and Bayesian networks to preserve privacy without compromising the quality of the data. • Build a robust framework to assess quality and confidentiality: univariate statistics, correlation matrix, survival analysis, differential privacy and several scores for inference and linkability.

R&D - Working on NLP & Computer vision topics Main Topics : • Created a named entity recognition (NER) function tailored to biomedical lexical field with dedicated API • Developed an internal application to streamline clinical data access for users, employing ETL and NLP techniques. • Designed a knowledge graph based recommender system with Neo4j & ElasticSearch to improve user experience. Cross-functional topics : • Conducted analysis on early oncologie studies using both univariate and multivariate statistics (PCA, FAMD, MCA) • Use of OCR and OpenCV to detect and analyze study design charts in research protocols • Automated various repetitive tasks using Selenium

Active Cosmetics & Professional Products Div - Supply & Data Excellence • Developed and managed data analysis tools for the Business Intelligence team, using Python, SQL and PowerBI • Initiated data-driven processes, enhancing team efficiency with a gain of ∼17 hours/week in productivity. • Built, monitored, and updated accurate KPIs and weekly reports for S&OP staff. • Conducted data analysis and created dashboards for risk assessment and out-of-stock scenarios. • Automated subsidiary order check-up, conducted data analysis, and created dashboards.

Student Job