About
Models, methods, and working systems.
I’m Xinyu Wang, a Ph.D. candidate in Computer Science and Engineering at the University of Connecticut.
My work spans representation learning, foundation-model adaptation, language agents, and ML systems. I enjoy problems that need both an algorithmic idea and the engineering to make it useful.
Two complementary kinds of work
As a Graduate Research Assistant in UConn’s Laboratory of Machine Learning and Health Informatics, I develop and evaluate learning methods for language agents, model merging, and structured generation. My research connects questions about representations and decision-making to concrete modeling problems.
As a Machine Learning Systems Engineer (Graduate Technician) with UConn’s Outage Prediction Model, I work on the engineering around operational forecasting workflows: data preparation, model execution, automated delivery, performance tuning, and recovery when inputs or jobs fail.
The common thread is taking responsibility for both the method and its behavior in a working pipeline.
Background
- Ph.D., Computer Science and Engineering, University of Connecticut — expected October 2026.
- B.S.E., Computer Science and Engineering (Honors), and B.A., Mathematics, University of Connecticut — 2019.
My technical toolkit includes Python, PyTorch, C++, SQL/PostgreSQL, distributed inference, Linux, MPI, and Slurm.
What I’m looking for
Full-time machine learning, applied science, and research engineering opportunities. I’m especially interested in LLMs, representation learning, and systems that bring learning methods into practical use.
Based in Storrs, Connecticut; considering opportunities in Connecticut, Boston, New York, and US-remote teams.
Email me · LinkedIn · Google Scholar
Away from work, there’s Hikaru.