Marc Chevallier

R&D and innovation management, including project planning, budgeting, resource allocation, and data team leadership. CIR tax credit documentation. Development of data correction pipelines, geolocation solutions (vector search, language model embeddings), and ETL pipelines (Mage.ai, Dremio). Scientific publications and international conferences. Recruitment, skills development, and internal projects (training, blog, Dataasso.fr). BI Project Manager for RATP Infrastructures.

Delivered 20 hours of Python instruction at the École nationale des chartes (PSL), 2022–2023 cohort.

“Machine Learning for Data Profiling” (CIFRE PhD Thesis) · Supervision of Master’s students (M1/M2) · University Teaching

Government database maintenance and data quality.

Development of a web scraping bot to enrich an existing database. Natural Language Processing (NLP) applied to scraped data, including Latent Dirichlet Allocation (LDA), TF-IDF, and Named Entity Recognition (NER). Web scraping to build a complementary database.

Shadow detection and extraction in high-resolution SAR imagery. Literature review and state-of-the-art analysis, selection of the most suitable segmentation algorithm (DOI: 10.1109/34.809108), and integration into the existing image processing pipeline. Creation of a database of extracted shadows. Registration of the entire image dataset using the segmentation algorithm. Extraction of invariant features to improve classification algorithm performance.

Machine Learning for Data Profiling | LIPN Laboratory