The use, complexity, and sophistication of Machine Learning is increasing rapidly within the High Energy Physics community. New ML solutions are being introduced to a range of problems, from simulation, to analysis, and showing improved performance over traditional methods. Although foundation models look promising for the future, for now bespoke model architectures trained with refined datasets are the predominant modes of development. In this seminar, three applications of ML in the HEP community will be presented. We will cover different architectures and training methods, looking at why they are suitable for each problem and the impact they have on Physics performance.