Statewide HR & Payroll Modernization

Government payroll is the definition of a system that cannot fail: every state employee, every pay period, under rules that have accumulated for decades on mainframe infrastructure. Modernizing it is digital government at its least glamorous and most consequential.

The problem

A statewide HR and payroll system ran on mainframe technology that was expensive to operate, hard to staff for, and difficult to extend. It needed a complete refactor to cloud-enabled infrastructure without disrupting the paychecks it exists to deliver.

The approach

Serving as lead data scientist on the refactor: mapping the legacy system’s data and rules, designing the modern data architecture that replaces it, and applying modern data principles (pipelines, lineage, measurement) so the new system is not just hosted differently but genuinely more observable and maintainable.

The outcome

[In progress]

Lessons Learned (so far)

There’s a lot of knowledge we are losing with the loss of some of the mainframe generations. These developers are insanely good at what they do. The amount of times I heard engineers refer to specific memory addresses and know exactly what’s happening at those addresses continues to blow my mind. With the onset of the AI generation of developers, I’ve very much enjoyed working in the juxtaposition of a generation that required physical manuals to work on their systems.