Transaction Risking DataOps

The Covid-19 pandemic has caused an unprecedented surge in demand on the welfare system, resulting in a hugely increased operational workload across DWP. To process benefit claims and pay customers their entitlement, there was an easement of controls that led to increasing levels of fraud and error. This is now at unprecedented levels, with £8.4bn overpaid in 2020/21. The Integrated Risk and Intelligence Service (IRIS) provides a service to support identification of singleton and organised fraud and error. IRIS has been challenged to be more real time and adapt to changing fraud risks. To deliver a better service, IRIS needs to: - Accelerate fraud and error prevention at the point of application and in advance of payment to reduce benefit overpayment and lower the debt burden on claimants. - Enhance DWP's fraud and error decision making through consolidating sources of risk into a single view of risk to inform routing and intervention types. - Better target operational resource to focus on the highest risk cases associated with the largest financial impact, enabling resourcing to risk. This is being achieved through investment in a cloud-based analytics platform where new fraud and error identification capabilities will be developed including machine learning, and a risk engine that will act as a funnel for all the different risk types and provide a triage function.

The project will deliver a core data integration capability that:

- Implements event-driven ETL pipelines and self-serve tooling to provide re-useable and extensible data workflows by integrating data from currently available systems and lines of business as well as securely ingesting and integrating 3rd party data (e.g. through automated data pipelines or by setting up API's). Some of these systems may not yet have the ability to interface with Transaction Risking and the infrastructure to support integration may not yet be in place. The Department is driving towards real-time analytics and so data should be incorporated in near real time where that is possible.

- Implements a data treatment service with obfuscation of personal data and security controls in place to ensure individuals and tools have appropriate data visibility, while retaining an appropriate level of detail so that the data can be linked and meaningful analysis of the data is still possible.

- Implements a data matching capability to resolve common entities across all data sets to enable claimants (and other entities of interest) to be tracked across multiple data sources. This will include creating the ability to link personal data securely in a cloud environment.

- Conforms to a consistent data model so data sets from multiple source systems are interoperable, making analytics development and insight generation more efficient.

- Is extensible to enable further integration of other event data sources in future e.g. other lines of business or data sources.