Project Leads: Jason Civjan, Zahra Taher, Adelin Ma, and Maeve Seyer, UW IT

Data Science Lead: Curtis Atkisson

UW-IT’s Student Innovation Lab (SIL) is dedicated to helping UW students from all disciplines explore solutions to problems facing the UW community. We provide access to UW data, resources, and consultations to support fast, effective, and responsible prototyping. One persistent barrier to student innovation has been access to realistic university data for projects, such as course schedules and registration records. This data lives in production systems that require UW approval and have serious privacy obligations. Without a safe alternative, students have resorted to web scraping, other risky workarounds, or simply abandoning their ideas. 

HuskyFetch is a student-built API that will provide access to non-sensitive UW data for prototyping solution ideas. The core data science challenge behind the development of HuskyFetch lies in how to make that data both safe and useful. Because this data touches student records, FERPA compliance is a foundational requirement. The approach is to generate synthetic student data using machine learning models that are trained on the real student data. The team is currently evaluating synthesizer models based on statistical realism (the synthetic data reflects true patterns) and privacy guarantees (nothing in the output can be traced back to a real student). Alongside this, the team is developing a probabilistic registration algorithm that enrolls these generated students into courses based on modeled likelihoods derived from the real data.