Abstract: Model registries are a key tool in addressing challenges around the ML lifecycle of models. They allow you to register, version, and manage models and their associated information throughout the deployment lifecycle. This session will go over MLOps challenges solved by model registry, what core requirements your team should think about when implementing one, and why a GitOps-based approach leads to the fastest time-to-market delivery of your ML models into production apps and services. Attend this session to learn: What an ML model registry is and what problems it solves What considerations to have when implementing a model registry Why a Git-based model registry will make both your MLOps and DevOps teams happy
Bio: Dmitry Petrov is an ex-Data Scientist at Microsoft with Ph.D. in Computer Science and active open source contributor. He has written and open sourced the first version of DVC.org – machine learning workflow management tool. Also he implemented Wavelet-based image hashing algorithm (wHash) in open source library ImageHash for Python. Now Dmitry is working on tools for machine learning and ML workflow management as a co-founder and CEO of Iterative in San Francisco.