Marwa Ezzouhri

Optimization of Simulation Models and Advanced Neural Networks (C++ - Python ) Developed machine learning models to predict and optimize parameters in real-time physics-based tire simulations, supporting industrial R&D workflows. Implemented differentiable simulation techniques (LLVM/Enzyme) to enable gradient-based optimization for control and design tasks. Designed neural networks to approximate complex physical equations, bridging physics-informed AI and industrial applications. Applied NLP/LLM-based methods to improve interpretability and usability of simulation models for engineers. Collaborated with cross-functional teams to integrate AI solutions into manufacturing pipelines and decision-making processes.

Data Processing and AI Development (Python) Built predictive analytics pipelines for large-scale heterogeneous data in software engineering and business contexts. Leveraged transformers and large language models (LLMs) to generate annotated datasets and train domain-specific assistants. Implemented anomaly detection and quality checks to support decision-making and improve software quality. Prototyped and benchmarked ML models, ensuring robustness, scalability, and usability for end-users.