What we work on.

We build and evaluate AI systems that take on software engineering work: writing and reviewing code, running agentic development workflows, and predicting quality attributes of software services. A model that returns a plausible answer is not the same as one you can put in a pipeline, so most of the effort goes into finding out which is which.

An AI tool only helps if people use it and can tell when to overrule it. We study how individual developers and whole organizations bring AI into their practice: what drives adoption, where trust breaks down, and which frictions leave capable tools sitting unused. What we find there decides what we build next.

Deciding what a system should do is still the hardest part of building one. The evidence for those decisions is spread across app store reviews, forums, support tickets, video platforms, and the memory of whoever was in the room. We build methods to find that evidence and reason about it, and we study how requirements work is really done in industry: in remote and hybrid teams, across software ecosystems, and under continuous delivery.

Privacy is where most of this work sits: what the GDPR asks of a team that ships several times a day, and what compliance looks like inside a small organization with no legal department. The rest of the area covers the other properties software gets judged on but nobody files a feature ticket for. Non-functional requirements, quality assurance for AI models, and technical debt are all settled by default unless someone makes them explicit.

Much of our work runs on large, noisy, human-generated data: reviews, traces, discussions, telemetry. We build the pipelines that analyze it, and we also study the analysis. Where can a model take over part of a qualitative study, and where do its interpretive steps let bias in?

We design and study courses that put students on real projects with real stakeholders: large multi-team capstones, community-engaged learning, and courses that pair AI skills with starting a venture. We report what worked, what did not, and what it takes to keep these experiences open to everyone. Much of this runs through OU's Software Studio, the engineering and entrepreneurship program Dr. Li co-chairs.