Roumen Vesselinov, PhD
Associate Professor, University of Maryland, Baltimore
I am a Lecturer at Johns Hopkins University and an Associate Professor at the University of Maryland, Baltimore. I am a researcher with more than 25 years of experience and more than 60 publications in peer-reviewed journals. My current work centers on artificial intelligence and machine learning — including prompt engineering and AI agents for research and analysis — and how these tools can accelerate empirical work in economics, governance, and policy.
How I can help with your research
I would be glad to advise and collaborate with students on their applied research projects, Capstone projects, and dissertations. My aim is to help you use modern AI and machine-learning tools to do better empirical research — faster, more rigorously, and more transparently.
- Framing a research question and matching it to the right method — regression, classification, forecasting, or text analysis.
- Using AI and large language models (LLM) responsibly for data analysis, literature synthesis, and turning documents into structured data.
- Prompt engineering and AI agents that automate parts of the research workflow while keeping every result verifiable.
- Machine-learning modeling on real data: decision trees, random forests, and neural networks, with honest evaluation.
Teaching at Johns Hopkins
I have taught the following courses at JHU:
- Big Data Analytics
- Statistics
- Macroeconometrics (Time Series Analysis)
- Macroeconomic Forecasting (Time Series Analysis)
Recent seminar — AI & Machine Learning for Governance and Policy Research
Most recently (August 6&7, 2026), I designed and led an applied seminar at SAIS for visiting fellows, “AI & Machine Learning for Governance & Policy Research,” a hands-on program in which participants run real analyses entirely in the browser — no coding or installation. The workflow is simple and rigorous: upload the data, ask in plain English, interpret the output, and verify every claim.
The seminar covers machine learning on governance data (decision trees, random forests, and neural networks) and text-as-data methods (dictionary sentiment and large language models) applied to real Federal Reserve, IMF, and World Bank documents. The full materials — datasets, policy documents, slides, and ready-to-run labs — are online at the site:
Research & specialization
Over more than two decades I have published 60+ peer-reviewed articles spanning applied statistics, econometrics, and machine learning. My specialization is in AI and machine learning, prompt-engineering analysis, and AI agents for research and analysis — with a practical focus on making advanced methods accessible to researchers who are not programmers.
A little beyond the research
Outside of teaching and research, I enjoy connecting technical ideas to real-world policy questions, and I like helping students find the version of a project that is both achievable and genuinely interesting to them. If you are curious about AI tools for your own work, come with questions — that is the best way to start a conversation.