Machine learning has the power to transform organizations in a way that few technologies have before. To capitalize on the automation of business-critical systems, companies use a variety of services to build ML systems and workflows - most of which reside in their cloud environment due to the size and sensitivity of data and models. Amazon Web Services (AWS) is a popular choice for companies looking to discover, build, and deploy ML solutions. However, regardless of the systems used, ML adoption carries business risk.
As such, we’re excited to announce the listing of Robust Intelligence on the AWS Marketplace. This enables AWS customers worldwide to proactively eliminate model failure at every stage of the ML lifecycle, from pre-deployment vulnerability detection and remediation to post-deployment monitoring and protection. By procuring the Robust Intelligence Platform from the Marketplace, customers can more easily instill integrity in their ML systems.
It’s simple for data scientists and ML engineers to get started with Robust Intelligence through the Marketplace listing. Prospective customers can purchase our SaaS-based service directly through AWS by swiping a credit card, adding it to their monthly bill, or using existing credits. The deployment keeps customers’ sensitive data and models in their own AWS environment, only sending test results to the Robust Intelligence control plane. Integrating an end-to-end solution to mitigate AI risk has benefits far beyond the data science organization, extending to compliance teams, executives, and ultimately the entire company.
Robust Intelligence proactively eliminates operational, security, and compliance risks at every stage of the model lifecycle and integrates with existing CI/CD workflows.
During development, models are validated for “production readiness” through AI Stress Testing which automatically detects and remediates vulnerabilities.
The moment a model is deployed to production, it is wrapped with an AI Firewall for real-time protection against anomalous data entering the system.
AI Continuous Testing allows companies to monitor the behavior of models in production to identify issues and automate root cause analysis of model failure.
Vendor security and compliance
Just as the adoption of ML introduces risk, the use of third-party tools and systems also carries certain risks. Hence today, we are very excited to announce that we've received our SOC 2 Type II certification, reaffirming our commitment to security and privacy to all our customers.
For our customers, this means they can trust that we take security and privacy of their data and models very seriously. To achieve such a high standard, we've worked closely with experts on application security assessments, security architecture reviews, and network penetration testing. We’ve hired top talent with security know-how, made sure our processes follow industry best-practice, and introduced tests and gatekeepers to quickly identify and block any security issues in development and production.
What's in the SOC 2 report?
Essentially, our SOC 2 report confirms that Robust Intelligence hits the highest industry standards when it comes to keeping information safe. The report was authored by an independent SOC 2 auditor that inspected our system and included their findings detailed explanations of the following:
- Strong authentication controls and limited access to data — Access to any customer data and model is strictly limited and monitored to those who need it to support customers.
- Continuous controls monitoring and incident response — We run continuous compliance monitoring to ensure that the key configurations our controls rely upon are in place, and that we're able to quickly respond to any issues that may arise.
- Employee security awareness — We run background checks prior to hiring and provide security training for all employees during onboarding and on an ongoing basis.
But none of this would be possible without the incredible team at Robust Intelligence. The journey to SOC 2 Type II compliance is long and takes a concerted team effort. Many individuals have worked tirelessly to help achieve this certification. And although it’s an important milestone, security and privacy is a continuous process. We’ll continue to work on additional certifications that are important to our customers, such as HIPAA, GDPR, and NIST 800-197.