Yahya Rochdi

Comparative study between morpho-geometric segmentation algorithms and deep learning models (ResNet, DenseNet) for vascular volume quantification. • Contribution to a longitudinal MSCT study analyzing pulmonary vascular remodeling in patients with pulmonary fibrosis. • Longitudinal analysis of vascular biomarkers (BV5/BV20, vessel caliber distribution) and correlation with disease progression (fibrosis volume, LIKERT score, interstitial lung disease extent)

Development of a document processing pipeline for RAG indexing in Azure DevOps, including text extraction, image description generation via VLM, cleaning, and conversion to semantic embeddings of Medical guides • Integration of the architecture with two user interfaces : one using OpenWebUI and the other using Flask • Evaluation and optimization of the semantic processing pipeline by testing different combinations of embeddings and rerankers (BM25, TFIDF, Matryoshka, dual encoders, cross encoders, LLM reranking), with performance • Development of a router to automatically classify medical images by organ, following exploratory analysis of DICOM tags (body parts, protocols, convolution kernel, etc.) using an LLM-based approach.

• Development of an advanced GEN-AI RAG (Retrieval Augmented Generation) application with a corrective feedback loop based on an LLM chatbot for the retrieval of relevant Project Management frameworks information, using Pinecone as a vector database and LlamaIndex then evaluated by RAGAS and Truelens in an Azure environment. • Comparison between different embedding models to ensure a vector representation with semantic meaning, followed by dimensional normalization to reduce retrieval time.

Deployment of a MeshSeg Net model that segment teeth and precisely label them on digitalized 3D dental surface models acquired by Intra-oral scanners on the web using AWS Lambda and API Gateway https://teethseg.vercel.app/

• Creation of an employee management application with Django and its deployment on the web and on a local server. • Design of a corporate Knowledge Management System project for managing professional files using SharePoint. • Learning various DevOps concepts and tools such as Dynatrace, Azure, AWS, and Jenkins for CI/CD.