MLSys-NYU-2022

Why we included this project

What makes this repo worth a look is that it comes from a graduate Machine Learning in Finance course at NYU Tandon that was built around real industry practice, not just theory. The 14 weeks each ship with their own folder of slides, runnable scripts, notebooks, and a requirements file, so a self-learner can work through the syllabus in order and reproduce the examples in a clean environment. The content is concrete: fraud detection and recommender systems appear as actual finance use cases, and the tooling is what working teams reach for, including Metaflow for pipeline orchestration, Comet for experiment tracking, Streamlit for sharing results, and Flask for serving predictions. For students, junior data scientists, or anyone trying to move past textbook ML, the week-by-week structure and the emphasis on coding habits, evaluation, and deployment make this a practical study path rather than a slide deck.