Overview
I take undergraduates into my research group and give them real problems. Not literature summaries or data cleaning for someone else’s paper — a question nobody has answered yet, scoped so that a motivated student can make progress on it in a semester or two.
I was an undergraduate at BYU myself, and the mentorship I got here is a large part of why I came back. BYU students are talented and genuinely hungry to do something worthwhile, and I intend to build my group around that. Some projects begin by pushing an established method until it breaks; others call for building something that does not exist yet, and I would like to develop those methods together. Either way the results genuinely determine what the group does next.
Part of what I want to offer is a clearer sense of what is actually within reach. Having worked at several world-class institutions, I can tell you which opportunities are realistic and what it takes to get there.
What you will do
- Own a question. You get a specific, answerable question and the latitude to pursue it. I will tell you when I think you are heading somewhere unproductive, but the project is yours to run.
- Build the instrument. Most projects begin by constructing the environment or harness that makes a question measurable. This is the part students underestimate and the part that teaches the most.
- Read deliberately. Two or three papers properly, not twenty superficially. We discuss them.
- Report honestly. Negative results are results. A well-run experiment that kills a hypothesis has done its job, and I would rather learn that in week three than month six.
- Write it up. Every project ends in writing, whether or not it becomes a paper.
How I mentor
I meet with research students weekly, one-on-one. Between meetings you should expect to be unstuck enough to make progress and stuck often enough to have something to bring.
My general approach: give you the smallest problem that is still genuinely open, let you struggle with it long enough to develop judgment, and intervene before the struggle becomes discouraging. Getting that timing right is most of the job. Tell me when I have it wrong.
Most of what we do in those meetings is get concrete about the problem — what exactly are you claiming, what would show it false, and what would you have to measure to find out. Learning to ask that of your own ideas is the real skill, and it transfers to everything else you do.
I will also be explicit about how research actually goes — that most experiments fail, that the first formulation of a question is usually the wrong one, and that revising it is progress rather than backtracking.
What I look for
You do not need prior research experience. You do need:
- Python fluency. You will be reading and modifying code from day one.
- Comfort with the math you have. Linear algebra and probability at the level of the standard courses. Gaps are workable; unwillingness to close them is not.
- Persistence. Research is mostly things not working. The students who do well are the ones who find that interesting rather than demoralizing.
- Honesty about results. Report what happened, including when it undercuts the hypothesis.
- Curiosity that is yours. The best reason to do research is that you want to know the answer. Résumé-building alone tends to run out partway through a hard project.
Coursework in machine learning or RL helps but is not required. If you have taken CS 401R, that is a good signal, and if you are taking it now, come talk to me before it ends.
Getting started
Email me with a short note covering what you have taken, what you are curious about, and roughly how many hours a week you can commit. A specific paragraph about a problem that interests you is far more useful than a polished résumé.
If you are earlier on and not sure whether research suits you, say so — that is a reasonable thing to find out, and a semester on a small project is a cheap way to learn it.
Credit and funding
Projects can be arranged for research credit, and funded positions are possible depending on the project and available support. Raise it when we talk; the arrangement should not be what determines whether you can participate.