Jurisdiction-as-Data
- regulatory precision vs. platform generality
- certification scope vs. jurisdictional coverage
- regulatory velocity vs. engineering velocity
- auditability vs. system complexity
- schema expressiveness vs. schema stability
How do you design, build, and maintain software when the constraints are regulatory, cross-organizational, and always changing? Four lines of work take up that question, ordered here from the most established to the newest.
Our work is grounded in evidence-based software engineering, design science, and method engineering (building artifacts, then studying them) and in the pattern-language tradition that runs from Christopher Alexander through the PLoP community. Patterns are not metaphor here. They are the established way practitioners name a recurring problem, the forces that pull against each other, and the solution that holds them in balance, and they stay legible to engineers and reviewers alike.
The dissertation line. Formalizing API-first design from a loose industry slogan into a structured, repeatable methodology: a survey of the state of practice, a systematic mapping of the literature, a taxonomy of the design space, and a process model teams can follow.
The differentiator. Gaming, fintech, and real estate share a class of architectural problems (per-jurisdiction compliance, document-mediated handoffs across organizations, partner extensibility) that recur as a small set of design patterns, drawn from three decades of production systems and written up as a pattern language.
Moving fast. AI agents are reshaping how software gets built and how research gets done: two related phenomena worth studying empirically rather than hyping, with live evidence from building real systems alongside agents.
Where the leverage is. The biggest return from AI agents is not in writing code; it is in operations: the document-mediated, rules-bound workflows that run regulated businesses, from underwriting and settlement to compliance and servicing. This line asks how to put agents into that work while keeping it correct, auditable, and jurisdiction-aware, the standards the domain already demands of people. It draws on production experience automating regulated document decisions in lending, and it is where the regulated-domain and agentic threads meet.