SARAH – Enterprise Scaling made Easy

The AI hype has been with us for years, but now AI is here to stay. Data should provide value – fast. The time spent on just testing AI via PoCs (Proof-of-Concepts) and MVPs (Minimum Viable Products) or similar initiatives are in the past.

Our experience is that companies and organizations struggle with new challenges after establishing for instance a Machine Learning model into production. Where should the data come from? How do we solve real-time challenges? Who validates that the models perform as they should? How do we re-train, and who owns the whole process? SARAH provides answers to all these questions. 

With SARAH you instantly get a full-scale environment with all imaginable functions that normally takes thousands of hours and long lead times to establish.

The benefits of SARAH

• Common approach and toolbox for a scalable Enterprise environment
• You can start small and scale quickly
• No lock-in effect – flexible architecture in Azure
• All services are hosted in private networks and isolated instances per project team (not exposed to the internet) to maximize security
• Easy to follow-up costs per project
• One version of the truth through a Master Data Lake
• Built-in rights management to data
• All components of the framework are released into production in Azure (GA; General Availability)

In addition to the benefits of quickly setting-up different environments (landing zones), SARAH includes a pre-configured setup of everything you could possibly need for a full-scale Data & AI solution. SARAH can also increase efficiency and reduce costs, enable new business models and faster time-to-value when it comes to analytical models.

The scaling technology

With code, you can set-up development, testing, and production infrastructure for various initiatives, by using ready-to-use scripts. No manual work is required, your infrastructure is set up automatically in Azure. You only pay for the Azure consumption you're using, and when a project isn't needed anymore, you can safely bring it down and know that all the code is left to re-script the entire setup again.

Here are further details on how SARAH gives you an accomplished Data & AI platform. For example, best practice for a modern data platform with lambda architecture, is implemented, including Auto-Lake. The MLOps framework allows you to create a pipeline to train a Machine Learning model with only two lines of code, something that typically takes close to 2000 lines of code. Of course, well-proven frameworks such as CAF (Cloud Adoption Framework) and WAF (Well-Architected Framework) are also implemented which provides you with a safe, stable, and secure environment. The next generation of data platforms with Data Mesh architecture are also built-in, through support via a 'Share Back' function that enables various initiatives/projects to subscribe to data sources created by other parts of your organization.

Combined expertise with big muscles

SARAH is the result of a close collaboration between Sogeti, Microsoft and the team behind the Open Source Framework ESML (Enterprise Scale Machine Learning). With ESML as a foundation, we have added additional functionality based on our knowledge and experience of large-scale Data & AI solutions in Azure. In short, SARAH is the result of all available best practices and years of experience in how to best create an Enterprise environment for DataOps, MLOps, and AI Factory in Azure.

With SARAH, we can now accelerate your entire set of technical prerequisites for Data & AI, in the shortest possible time. Get in touch with us and we will be happy to tell you more about how SARAH can help you and your business.