prospective students

(Last update: Jul. 1, 2026)

Welcome

I am recruiting students who want to build rigorous and usable causal inference methods. Our lab’s research program is Pragmatic Causal Inference: making causal questions as usable as prediction workflows while keeping assumptions, uncertainty, and validity explicit.

A typical data scientist can run a prediction model with model.fit(X, Y) and model.predict(new_X). Causal questions are harder because they ask what would happen under an intervention. That requires assumptions, identification, estimation, diagnostics, and decision-facing interpretation. Our lab works on making that workflow more reliable and more usable.

Research Directions

  1. Partial Identification: We develop methods that return valid bounds or sensitivity analyses when a point estimate is not justified.

  2. Practical Causal Learning: We study estimators and workflows under mild, defendable assumptions such as front-door, proxy, mediator, or overlap structure.

  3. Amortized Causal Inference: We study simulation, pretraining, and reusable causal machinery that reduces repeated graph, identification, and estimation burden.

  4. Real-World Deployment: We work on causal systems for biomedical digital twins, health decisions, and other settings where causal answers must support action.

How We Work

Good research in this lab means being precise about assumptions, honest about uncertainty, and persistent about hard technical problems. I value students who can take ownership, communicate clearly, receive feedback seriously, and treat collaborators with respect.

I do not expect incoming students to know every area already. I do expect a willingness to build the needed foundations in probability, statistics, machine learning, programming, and causal inference.

Who Might Be a Good Fit

You may be a good fit if you are excited by questions like:

  • When is a causal effect identifiable from the available data and assumptions?
  • What can we still say when point identification fails?
  • How can modern machine learning help without hiding causal assumptions?
  • How can causal methods support real decisions in health, policy, science, or AI systems?

Strong preparation in math, statistics, machine learning, or systems is helpful. Curiosity and discipline matter at least as much as having already worked in causal inference.

Getting Involved

Prospective PhD students should apply through the UIUC School of Information Sciences PhD program. Current UIUC students may email me with a short description of their background, research interests, and what they hope to work on.

When emailing, include:

  • your CV or resume;
  • a brief description of your technical background;
  • one or two research questions from this page that genuinely interest you;
  • any relevant papers, projects, or code.