Gabin Maxime Nguegnang

* Trained advanced AI models for specialized, real-world problems, aligning outputs with complex user intent and domain needs * Transformed mathematical concepts into accurate ground-truth data for advanced reasoning tasks * Developed rubric criteria that standardized quality assessment, resulting in improvements in the model performance * Collaborated with cross-functional teams, providing actionable insights and feedback loops that drove continuous improvement in production environments

I study optimization methods for training deep neural networks, developing strategies for parameter initialization and learning rate selection. My work offers theoretical and practical insights into the development of deep learning, validated through experiments on neural networks with Gaussian data. I have experience in model training with Python, PyTorch, and HPC systems, translating AI research into practical solutions. I also teach exercise classes on Machine Learning, Deep Learning, and Optimization.

I studied deterministic algorithms for training deep neural networks, establishing convergence properties of gradient descent for linear networks with constant and adaptive learning rates. My work advances optimization techniques for deep learning and supports effective model development. I also taught tutorials on Continuous Optimization, Mathematics of Data Science higher mathematics for engineer students.

I Applied Machine Learning techniques to detect anomalies in fuel consumption data, optimize consumption, and limit fuel pilferage in telecommunication base stations. I trained Machine Learning models (including Neural Networks) that contributed to the security of approximately 84617 liters of fuel. Additionally, deployed predictive Machine Learning models into web applications, which facilitate visualization and monitoring of fuel consumption at each base station monthly