Productionizing Your ML Code Seamlessly

Explore a comprehensive talk from EuroPython 2018 on seamlessly productionizing machine learning code. Learn about the challenges of taking experimental data science and ML projects from notebooks to production environments. Discover strategies for implementing regular retraining schedules and maintaining long-term reliability. Delve into two main themes: understanding what running an ML model in production entails and improving development workflows to streamline the path to production. Gain insights from real-world examples at Yelp, including migrating a pandas/sklearn classification project to production using pyspark. Acquire framework-agnostic advice applicable to listeners from various backgrounds on topics such as data sources, prediction, monitoring pipelines, designing for change, and quality assurance.