Resolver Transforms Operations with Machine Learning

The Customer

Resolver, a Kroll business, isa specialist in risk intelligence, safeguarding over $6.5 trillion in market cap for more than 1,000 global companies.

Our Role

We supported Resolver through building a platform to onboard open source large language models to augment their AI workflows. This included rapid adoption of open source Large Language Model and democratising the prompt design to enable business teams to experiment more efficiently.

The Challenge

Resolver needed to increase automation and efficiency of data classification to find risks in text and images, in order to provide faster and more accurate risk detection analysis.

The current setup did not fully meet their needs for flexibility, customisation and cost management. In addition, the models needed to integrate seamlessly with Resolver’s own core platform technology.

"Working in partnership with Qodea, we rapidly built repeatable MLOps processes, flexibility in selection and deployment in LLM models and business processes to ensure we meet our objectives of excellent service to our customers through speed of our delivery and effectiveness of our actions and insights. The insights that all parties contributed to the project were central to successful delivery. "

^{Kevin Fletcher - Chief Data Officer}

The Solution

Working in partnership, we quickly built repeatable Machine Learning Ops (MLOps) processes, delivered flexible large language models (LLM) and business processes that centre around Resolver delivering an excellent customer service.

The three key deliverables entailed:

Optimised LLM Model via effective prompts, standardising model sourcing and automating deployment processes.
Built infrastructure for MLOps to run automated environment builds via continuous integration and continuous deployment.
Built a generic platform for rapid adoption of open source large language models; providing Resolver with the flexibility and efficiency they required.