MLOps: streamlining machine learning for CSPs
12 June 2023 | Research
Strategy report | PPTX and PDF (7 slides) | AI and Data Platforms
"With MLOps, CSPs can optimise deployed ML models through monitoring and the automatic retraining of poorly performing models on the latest available data."
The term ‘MLOps’ is used to refer to both machine learning (ML) operations and DevOps for machine learning. It is a set of practices which brings together the data engineering and model development activities done by data scientists and the model deployment and maintenance activities done by operations teams (and/or ML engineers).
As the usage of ML grows, communications service providers (CSPs) will need more efficient and effective methods of creating ML models and bringing them into production. Scaling the deployment and operations of ML will require automation, better approaches to ML management and governance and increased monitoring of ML models in production environments.
Information included in this report
- A more-detailed definition of MLOps
- Insights into why CSPs need MLOps and the use cases and benefits that can be achieved from its usage
- Insights into the different considerations that must be made by CSPs attempting to implement MLOps.
Author
