Building an ML Experimentation Platform for Easy Reproducibility

Discover how to build a robust ML experimentation platform for easy reproducibility in this 36-minute conference talk from Data Council. Learn to leverage data versioning engines to intuitively version ML experiments and reproduce specific iterations. Follow along with a live code demonstration that covers creating a basic ML experimentation framework using lakeFS on Jupyter notebook, reproducing ML components from specific experiment iterations, and building an intuitive, zero-maintenance experiments infrastructure using common data engineering stacks and open-source tools. Gain insights from speaker Vino Duraisamy, a developer advocate at lakeFS with extensive experience in data management, batch processing, and MLOps from companies like NetApp, Nike, and Apple.