Thomas G.

Build a graph-oriented database from a complex business context : next-gen nuclear reactor. Direct link with C-level and exec in charge of information management. Work with GraphDB and Neo4j and a lot of RDF data.

Design and implement python routines including API calls, ETL methods and raw data processing. Start from scratch,

Conception and implementation of anomaly detection programs. Focus of litter,

refuse and waste in urban, semi-urban and natural environments. Use of deep

learning models and specific methods.

Working with small datasets made of images acquired by drones and fixed

cameras

I used the latest Yolo, DETR, Hugginfaces models (FastSAM, MobileSAM) and approaches using scikit-image, including structural similarity.

Leading data strategy and technology, building a scored database of

scientific and medical knowledge in immunotherapy. A founding member of the

Marseille Immunology Biocluster (MIB), which secured €97M in funding from France

2030.

Performed data mining on 10 million medical studies. Extracted

insights to enhance research accessibility and knowledge discovery. Designed a scientific article recommendation engine, creating a structured knowledge base of articles, authors, and research laboratories.Leading data strategy and technology, building a scored database of scientific and medical knowledge in immunotherapy. A founding member of the Marseille Immunology Biocluster (MIB), which secured €97M in funding from France 2030. Performed data mining on 10 million medical studies. Extracted insights to enhance research accessibility and knowledge discovery. Designed a scientific article recommendation engine, creating a structured knowledge base of articles, authors, and research laboratories.