© Hien Luu, Max Pumperla and Zhe Zhang 2024
H. Luu et al.MLOps with Rayhttps://doi.org/10.1007/979-8-8688-0376-5_6

6. ML Observability Infrastructure

Hien Luu1  , Max Pumperla2 and Zhe Zhang3
(1)
Santa Clara, CA, USA
(2)
Bad Segeberg, Germany
(3)
Sunnyvale, CA, USA
 

Congratulations! After weeks or months training, your model is finally complete and deployed to production, and integrated into your recommendation system. The A/B testing results show a meaningful and positive impact on business metrics. However, your co-worker reminds you that your work is not yet finished. Your model needs continuous monitoring to ensure its performance remains optimal. In other words, your model is at the beginning of its operational journey.

It is a well-known fact that ...

Get MLOps with Ray: Best Practices and Strategies for Adopting Machine Learning Operations now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.