System for Machine Learning Model Training and Usage

Machine learning and Data Science topics are worldwide trends right now. Many students learn them in universities, boot camps, and online courses. Moreover, they do that not only for the sake of learning but also for building projects and solving complex challenges. However, once the model is trained, it still requires a significant effort to serve it in production. Also, there is a notable growth of new AI startups globally. Specifically, early-stage startups. Unfortunately, they do not have enough resources for expensive AI cloud solutions by tech giants or big teams for developing their infrastructure for machine learning pipelines. This paper describes a system where students and early founders could train their models without running their training scripts on their machines and deploy them for inference seamlessly in one click, without any need to manage infrastructure. Index Terms: MLOps, machine learning, machine learning infrastructure. The paper contains 55 pages, 3 chapters, 25 illustrations, 5 appendix