Situation
Our client, one of the largest financial institutions in the Netherlands, is a multinational banking and financial services company that employs more than 40000 FTE worldwide. The client is in transition from legacy infrastructure to the cloud. During the transition, many ML DevOps teams are struggling with rapid experimentation and operationalisation of their machine learning projects. Our client was looking for a partner to enable them to operationalise machine learning solutions to timely deliver (additional) services to their customers.
Approach
We have taken the journey with our client to implement ML Ops. The implementation has been done on their on-premise servers, in such a way that a future transition to the cloud can be accelerated. Initialising the project with our VQD blueprint, we have been able to fully automate multiple machine learning processes e.g. serving predictions, training- and publishing jobs. These processes are all within the on-premise infrastructure (except for the code repository in the cloud).
Results
Our client is benefiting from rapid experimentation, allowing them for the first time to see and experience the competitive advantage machine learning can bring to the business. This with the ability to deploy any new experiments to production in 10 minutes.
