Join the Group

Thank you for your interest in the Minnesota NLP group. I am always looking for passionate, self-motivated students. I prefer to start working with students through my classes, directed studies, summer internships, or short-term projects, and to begin formal advising after that. Our group page and recent publications show what we work on.

How I advise

From my side, you can expect three things: honest and direct feedback, resources and freedom to focus once trust is built, and full support for whatever career path you choose, in academia, industry, or anything in between.

For PhD students, my involvement changes over time. Your autonomy follows a U-curve, and the impact I expect from you follows a J-curve.

Ground rules

• I give you problems a step beyond your current ability, because that is where growth happens. Your job is to break the problem down, detect risks early, and raise issues before they become failures.
• Autonomy scales with trust. Consistent progress earns freedom over topics, schedule, and where you work. If progress stalls or you go quiet, I step in with shorter check-ins and more concrete tasks until momentum comes back.
• Convince me with data. A small experiment, a table, and a cost estimate change my mind faster than a long argument.
• Science, not engineering. Building a system is not a contribution by itself. Every project has to answer one question: what do we know now that we didn't before?

What I look for

• Passion for AI/ML research and the persistence to push through setbacks. Prior quantitative research in another field is a plus.
• You can explain why your work matters and teach others what you learned, in papers, talks, and reports, including null results.
• You work well in a team, make data-driven decisions together, and are comfortable with collaborators from other disciplines. Know your gaps, and build a team that covers them.
• You believe the way to find a big-picture vision is to get into the details first.
• Consistency beats last-minute heroics. Make progress every week: a figure, an example, a proof-of-concept. Papers rushed in the final week always show it.
• Don't disappear. If you are stuck, say you are stuck. Silence is how trust is lost.

How to get involved

• Write in your own words. Form responses and emails written entirely by AI are ignored immediately. I read them to see how you think, so a short, specific message in your own voice beats a polished generic one.
• Current UMN students. Fill out the interest form for a lab internship. Master's and undergraduate interns commit at least 10 hours per week for at least 3 months. I review applications on a rolling basis at the end of each semester and interview those who pass the first round. Fewer than 15% are accepted, so describe your interests, strengths, weaknesses, and prior research experience in detail.
• Prospective PhD students outside UMN. Apply to the UMN CS&E graduate program and list me as a potential advisor. The admissions committee reviews applications and I am not directly involved, so I cannot reply to individual emails.
• Postdocs, visiting faculty, and visiting graduate students. Email me your CV and a short description of your research interests.
• PhD students advised by other faculty are not eligible for my advising or collaboration. Please ask your advisor to contact me first.
• Questions about our work? Email anytime.

Recommendation letters

Because of limited bandwidth, I write letters only for students who have done research with me: directed studies, research internships, thesis committees. If you took my NLP class, I can write one only if your course project has research novelty or can be extended to a conference or workshop submission. Let me know at least three weeks before the deadline; otherwise I cannot guarantee the letter will be on time.