University of Massachusetts

Building Applications around Diverse Campus Requirements

One of the biggest challenges in the University of Massachusetts system is developing technology that meets the distinct needs of each campus. Individual campuses have business processes they want to maintain, and those requirements can compete. We have to find commonality in how work gets done without making applications excessively complicated to manage and support.

Our business partners are central to that effort. IT is not equipped to determine how every administrative process should work. We rely on colleagues with that expertise to assess existing practices and guide campuses toward decisions that serve the collective system, and not just individual campus priorities.

Moving from Reactive Monitoring to Proactive Reliability

We have worked to become more proactive in how we monitor our systems. Instead of waiting for an alert before taking action, we are trying to identify problems earlier. Redundancy is also important because a single thread should not bring down an entire line of business operation.

Developing these capabilities requires people and technology. Automation helps us work more efficiently and gives us more capacity to improve redundancy, provide earlier notifications and remediate problems before they become operational problems.

Combining Testing Discipline with Flexible Delivery

Quality starts before an application reaches its users. I have stressed developer-led testing and test automation to catch problems within the development pipeline, before user acceptance testing. I have also come to appreciate the importance of documenting requirements and technical details. Conversations at the beginning may feel time-consuming, but unclear expectations can create problems later in development.

We do not use one development method across every team. Some teams work in true agile fashion, while others operate within waterfall-style project timelines and milestones. We still try to organize their work into smaller components and sprint-like increments, making substantial projects easier to tackle without requiring the entire program to follow an agile model.   

Automating Routine Work While Retaining Accountability

AI can identify code issues, surface document themes, support comparative reviews and assist with routine tasks like moving files and transferring data between systems through personal agents.

I use AI to assist developers and administrative teams, not replace their responsibilities. I don't want developers using AI to write entire applications because they must understand how those applications work to support, maintain and enhance them. Administrative outputs must also retain human oversight to ensure they are accurate and understandable.

Our hyperautomation program began primarily with UiPath RPA. Before selecting a process, we examine how manual it is, the number of source systems, the type of data handled, the people involved, its error rate and its criticality.

Having a smaller staff strength, we prioritize time-consuming processes with lower error rates, which reduces the time spent identifying and fixing errors. We also consider higher-error processes when errors are predictable, and the appropriate corrective actions are known. We have streamlined processes that previously took staff in the president's office days to complete and improved how information reaches campuses.

Making HR and Finance Transformation the Priority

Replacing our HR and finance ERP systems is my primary focus over the next few years. This will be a significant transformational effort for our system, and it will involve considerable discussion about business processes.

Modern systems cannot be customized as much as some legacy systems, which is both a blessing and a curse. Customization remains possible, but it takes more time to support and maintain systems and extends upgrade cycles. This is more than a technology replacement. It is also an opportunity to reconsider how work is performed across the university system.