We all love Pandas, Sklearn and the rest of the PyData stack. They allow us to conduct complex analysis and implement cutting-edge machine learning models simply and easily. However after the initial model fitting a common challenge often arises - how do we put these models in production ensuring that it fits into a larger organizational architecture? In this talk we outline the various strateg
The PyData stack offers a remarkably powerful toolkit for building complex machine learning and analytical components quickly. However, machine learning doesn't happen in a vacuum. It is part a large system of enterprise software responsible for data processing and must play- well with other tools in the ecosystem. In order to get the benefits of rapid development while not sacrificing the non-functional requirements, MaxPoint as implemented and tested multiple deployment models for software relying on the PyData stack. This talk we walk through these various deployment models and discuss the trade-offs of the approach.