– Roman Moser (Senior Consultant),
– Spyros Cavadias (Consultant)
MLOps is a set of practices to transition machine learning models from isolated experiments to reliable and maintainable production systems. It serves as a counterpart to the DevOps practice in classical software development and includes everything from data ingestion and preparation to model training, tuning and serving as well as monitoring and orchestration. In this talk we will present all steps of an MLOps workflow on a real-world example and show how to implement a production ML system with a set of open source tools. Furthermore, we present the different MLOps components and solutions of various cloud platforms.
- Why MLOps is needed
- ML lifecycle and MLOps Theory (CI, CT, CD)
- Overview of Big Data Provider Solutions (AWS, GCP, Azure)
- Walkthrough MLOps workflow (case study)
Everyone interested in machine learning and data science.
No prior knowledge or experience required.