Like most software engineers, data scientists generally enjoy working on complex problems. They want to create innovative solutions that bring value to their organization. Tools and processes deemed to be obstacles are often a source of frustration that take time away from designing, developing, testing, and promoting models into production. Data scientists would benefit most from processes that automatically run in the background rather than labor-intensive tasks that are difficult to streamline and bubble up to executive teams.

Conversely, data science leaders tasked with mitigating AI risk across the organization without stifling innovation need to ensure that certain standards are implemented and enforced. VPs and C-level executives similarly need aggregate information from their data science orgs that is accessible and easily understood.

Many large data science teams are adopting a hub and spoke structure, which refers to an organizational model with separated (spoke) teams that report to and receive direction from a central hub for data and policy information. This helps businesses divide responsibility between teams efficiently, and allows different spoke teams to focus on different problems. The hub can be optimized for these teams by having common access to tools and data, as well as standardized governance policies in place.

Robust Intelligence offers specific features to help streamline governance, reporting, and administration needs across data science organizations, enabling teams to work in different problem spaces and collaborate efficiently as needed. These features are built on top of our core pre-and post deployment model testing framework, which spans across three key risk categories: Operational, Security, and Fairness. Our platform also ensures security and privacy of information across teams. Below we highlight some of our key governance features:

Governance Dashboard

Each workspace comes with a governance dashboard where projects can be tracked against key metrics. The governance dashboard provides a central window into all models in production in a workspace. Health status and variance across time of business metrics are available for models in production to help evaluate model risk against business risk. This provides managers and executives with easily digestible information about each of their models and their associated risks. These dashboards make it easy to track model owners and take action on specific projects as needed. This feature is also useful for tracking key business KPIs related to each of the models in production.

Autogenerated Model Cards

The Robust Intelligence platform autogenerates model cards to simplify and streamline documentation of model validation and monitoring results for auditing, self-imposed regulation, and external regulatory reporting requirements. This feature makes it easier for governance, risk, and compliance teams to work together with data science teams to align technical insights with regulatory requirements, and standardizes model documentation across the organization to eliminate manual work for data scientists.

Workspaces

A workspace is an environment for organizing and accessing all of your Robust Intelligence assets. All of the Robust Intelligence assets in a workspace are unique to that workspace and serve as dedicated environments for specific teams. These dedicated workspaces enable organizations to use role-based access control (see below) to limit user account privileges to the users in that workspace. This protects sensitive information specific to certain data science teams from being shared across the organization. A workspace organizes a data science team’s model activity into projects. Projects are organized according to the machine learning problems they are aiming to solve.

Role-Based Access Control

Role-based Access Control (RBAC) is a mechanism that restricts permissions and enables user-privileges for authorized members on the RI platform. These restrictions are mediated by role, and protect sensitive information between and within workspaces. RBAC creates different permissions for different types of users, such as Admin members, Viewers, and VP users.

These governance features solve pain points for data science teams (spokes) and facilitate effective communication of model and business risks to organization leadership (hub). At Robust Intelligence, we strive to mitigate AI risk and instill model integrity. In doing so, we enable data science teams and facilitate governance across organizations to operate more efficiently and effectively.

Effective AI governance is one of many components that go into instilling integrity in ML systems. To learn more, request a product demo here.