About
Hi! I'm Jiwon Chang, a computer science PhD candidate at the University of Rochester. I research data management systems & machine learning, advised by Dr. Fatemeh Nargesian.
My research aims to bring the power and expressibility of modern ML models, from decision trees to LLMs, into SQL and relational data systems. I envision a system that is expressive, efficient, and explainable.
- Expressivity means combining the algebraic guarantees of SQL with the flexibility of natural langauge with LLMs.
- Efficiency means we want queries to take not much longer than a standard SQL query, despite involving sophisticated ML model inference.
- Explainability means that alongside the query result, the system explains what steps it took to answer that query, and what data was influential in giving that particular answer.
My interests outside of research include bug friends and houseplants, science fiction, gaming, minimal and intentional living.