Shreya Shankar • 11/29/2021

The Modern ML Monitoring Mess: Rethinking Streaming Evaluation (1/4)

This article, the first in a four-part series, analyzes the fragmented state of ML monitoring (MLOps). It argues that traditional evaluation metrics are insufficient for streaming ML deployments and proposes a rethinking of how to define and monitor for failures in production ML systems, especially in high-stakes applications.

0 comments

#DevOps #Mlop #Production ML