MARIEM MAKNI

- Explored the 3D TransUNet architecture, which enhances the nnUNet framework through the integration of transformers. - Utilized the Brats Met dataset for training and validated on external datasets to ensure robust model generalizability. - Explored various loss functions, notably integrating the Hausdorff distance metric to refine performance metrics.

Transcription and Online Session Control System Solution Engineer •Implemented advanced audio processing techniques to enhance transcription accuracy across different conditions. •Assessed and adapted open-source models for improved recognition and transcription of Arabic and English. •Employed Large Language Models (LLMs) using the GPT API for structured data extraction from transcriptions.

Machine Learning Operations (MLOps) Internship •Developed a scalable and robust backend infrastructure with FastAPI. •Performed comprehensive testing and evaluation of machine learning models using MLFlow and Sonarqube. •Orchestrated container deployment across Kubernetes clusters, enhancing resource management. •Engineered and implemented a CI/CD pipeline using GitLab to streamline development processes.