1. Data-centric AI
Measuring the value of data points, groups, and contributors while accounting for strategic behavior and real-world structure.
Statistics Ph.D. Candidate · The Ohio State University
Expected to graduate in May 2027
Statistical foundations for
reliable and accountable AI
Hi, I am Kiljae Lee, a Ph.D. candidate in Statistics at The Ohio State University, where I am advised by Professor Yuan Zhang. I develop principled and scalable methods for data-centric AI, explainable AI, LLM evaluation, and uncertainty quantification. My work connects statistical theory with practical algorithms for understanding data, model behavior, and structured human preferences.
I am on the 2026–27 job market. Feel free to reach out.
News
Recent updates
- Oct 2026
Selected for an ICSDS 2026 Student Travel Award from over 100 applicants! Declined to keep my invited talk.
- Oct 2026
Recognized as a NeurIPS 2026 Top Reviewer!
- Sep 2026
Generalized Priority-Aware Shapley Value was accepted to NeurIPS 2026!
- Aug 2026
Invited to speak at the IMS International Conference on Statistics and Data Science (ICSDS 2026) in Split, Croatia, in December 2026.
- May 2026
Invited to speak at the ASA Statistical Learning and Data Science Conference (SLDS 2026) in New York in November 2026.
- Apr 2026
Priority-Aware Shapley Value was accepted to ICML 2026!
Research focus
Making modern AI systems more reliable
My research studies how data, contributors, predictions, and preferences should be valued when structure and strategic behavior matter.
2. Explainable AI
Designing Shapley-based attribution methods that respect priority, precedence, and dependencies among contributors.
3. LLM evaluation
Modeling soft, cyclic, and multi-criterion preferences to evaluate modern generative AI systems more faithfully.
4. Uncertainty quantification
Building conformal prediction methods with statistical guarantees and substantially lower computational cost.
Selected work
Featured publications
Generalized Priority-Aware Shapley Value
Brings cyclic, soft, and multi-criterion human preferences into LLM evaluation by modeling structured preferences as directed weighted priority graphs.
Priority-Aware Shapley Value
Introduces PASV, a Shapley-based attribution framework that incorporates precedence constraints and contributor-specific priority weights for more structure-faithful data valuation and feature attribution.
Faithful Group Shapley Value
Develops a robust and efficient method for group-level data valuation that is resistant to strategic group splitting and supported by theoretical guarantees.
Leave-One-Out Stable Conformal Prediction
A computationally efficient conformal prediction method that leverages leave-one-out stability to improve uncertainty quantification for black-box models.
* Equal contribution
Recognition
Recent honors and service
Recognition for research contributions and peer review across the machine learning community.
Student Travel Award
IMS International Conference on Statistics and Data Science
Selected from over 100 applicants; declined under the conference's one-talk-per-person policy to keep my invited talk at ICSDS 2026.
NeurIPS Top Reviewer · ICML Gold Reviewer
NeurIPS 2026 · ICML 2026
Recognized as a NeurIPS 2026 Top Reviewer, and received the ICML Gold Reviewer Award for ranking among the top 25% of reviewers based on area-chair ratings.
Ransom & Marian Whitney Award for Research
Department of Statistics, The Ohio State University
One of two awardees, recognizing research independence, creativity, originality, and publication potential.
Best Paper Award Honorable Mention
ICML DataWorld Workshop
Awarded to Faithful Group Shapley Value; ranked among the top three of 55 accepted submissions.
Experience
Research grounded in practice
Alongside theoretical research, I work on statistical and machine learning problems in industry and interdisciplinary consulting.
Data Scientist Intern
United Airlines · Chicago, IL
First at United to model customer search data for commercial use, as a proxy for latent total demand; built demand forecasting models that extend willingness-to-pay estimation beyond observed bookings.
Graduate Research Assistant
The Ohio State University · Columbus, OH
Research with Yuan Zhang on data valuation, attribution, and uncertainty quantification for modern AI systems.
Statistical Consultant
The Ohio State University · Columbus, OH
Advised six clients across psychology, agricultural science, civil engineering, and related fields on statistical modeling and analysis.