MLOE-04: Establish ML roles and responsibilities
Understand the roles, responsibilities, ownership, and required interactions across teams to maximize overall effectiveness. An ML project typically consists of multiple roles, with defined tasks and responsibilities for each. In many cases, the separation of roles and responsibilities is not clear and there is overlap.
Implementation plan
-
Establish cross-functional teams with roles and responsibilities - Enterprises often struggle getting started with their first enterprise-grade ML platform. This is partly due to the ML platform architecture having many components. Complexities around data science, data management, and model and operational governance also contribute to this struggle. Building an enterprise ML platform requires the collaboration of different cross-functional teams. The different personas from the different teams should each have a different role and responsibilities. They all should contribute in the build-out, usage, and operation of an ML platform. Structure your ML organization to support your defined business outcomes. Examples of ML technical roles and responsibilities with their definitions include:
-
Domain expert - Has valuable functional knowledge and understanding of the environment that the ML problem must be framed in. Helps ML engineers and data scientists with developing and validating assumptions and hypotheses. Engages early in the ML lifecycle and stays in close contact with the engineering owners throughout the evaluation phase.
-
Data engineer - Transforms data into a consumable format for machine learning and data science analysis.
-
Data scientist - Employs machine learning, statistical modeling, and artificial intelligence to derive insights from the data.
-
ML engineer - Turns reference implementations of ML models developed by data scientists into production-ready software.
-
MLOps engineer - Builds and manages automation pipelines to operationalize the ML platform and ML pipelines for fully or partially automated CI/CD pipelines. These pipelines automate building Docker images, model training, and model deployment. MLOps engineers also have a role in overall platform governance such as data and model lineage, as well as infrastructure and model monitoring.
-
IT auditor - Responsible for analyzing system access activities, identifying anomalies and violations, preparing audit reports for audit findings, and recommending remediations.
-
Model risk manager - Responsible for ensuring machine learning models meet various external and internal control requirements. These requirements include: model inventory, model explainability, model performance monitoring, and model lifecycle management.
-
Cloud security engineer - Responsible for creating, configuring, and managing the cloud accounts, and the resources in the accounts. Works with other security functions, such as identity and access management, to set up the required users, roles, and policies to grant users and services permissions to perform various operations in the cloud accounts. On the governance front, cloud security engineer implements governance controls such as resource tagging, audit trail, and other preventive and detective controls to meet both internal requirements and external regulations.
-
-
Enable easy mechanisms to control access and grant permissions for various ML roles – Appropriate user access controls are essential for governance; they enable practitioners to access the tools they need to do their jobs, while ensuring data privacy and security.
-
Avoid the use of one-time methods for managing access policies for large teams which contain multiple roles for performing various ML activities, such as data preparation, training, and model monitoring.
-
Use Amazon SageMaker AI Role Manager to make it easier for administrators to control access and define permissions for users. Administrators can select and edit prebuilt templates based on various user roles and responsibilities. The tool then automatically creates the access policies with the necessary permissions within minutes, reducing the time and effort to onboard and manage users over time.
-
Documents
Blogs
