Explore symptom-based alerting for machine learning in this 25-minute conference talk from SREcon23 Europe/Middle East/Africa. Discover practical insights on implementing effective monitoring for ML stacks, going beyond traditional software monitoring practices. Learn which metrics to prioritize, how to detect issues in real-time, and whether existing tools suffice or if an MLOps platform is necessary. Gain valuable knowledge from the speaker's experience monitoring over 30 machine learning use cases, equipping you with strategies to enhance your ML monitoring capabilities.
Symptom-based Alerting for Machine Learning - Lessons from 30+ Use Cases
Overview
Syllabus
SREcon23 Europe/Middle East/Africa - Symptom-based Alerting for Machine Learning - What I Learned...
Taught by
USENIX
Related Courses
-
Machine Learning Monitoring in Production - Lessons Learned from 30+ Use Cases
-
Reliable Data for Large ML Models: Principles and Practices
-
Implementing Open-Source Observability at Maersk - SREcon23 Europe/Middle East/Africa
-
9 Things You Should Do When Starting to Use SLOs
-
Towards Observability for Machine Learning Pipelines
