I am an undergraduate student studying Joint Honours Mathematics and Computer Science at McGill University. I am broadly interested in statistical machine learning and theoretical computer science, with a focus on building distribution-free inference systems that are both mathematically rigorous and computationally robust.
Core Interests
- Conformal Prediction & Selection: Designing predictive inference methods with finite-sample error control, with a focus on candidate selection, False Discovery Rate (FDR) control, and post-selection validity.
- Distribution-Free Inference: Investigating non-exchangeable settings, weighted conformal frameworks, and robust multiple testing under covariate shift.
- Statistical Machine Learning: Exploring the interface between predictive hypothesis testing, e-values, and high-dimensional estimation.
Research
I am currently an undergraduate research assistant with Prof. Archer Yang (McGill/Mila). My research centers on conformal selection—developing statistical frameworks that select promising candidates from large pools (e.g., in molecular discovery and high-stakes decision making) with rigorous, finite-sample False Discovery Rate (FDR) guarantees.
While standard conformal prediction provides valid marginal coverage on average across an entire population, it fails on selected shortlists due to selection bias (the “winner’s curse”). My work explores constructing valid conformal p-values and e-values coupled with multiple testing procedures (such as the Benjamini-Hochberg procedure) to retain high statistical power without sacrificing finite-sample error bounds.
Below, you will find a growing digital archive of my technical notes, algorithm implementations, and mathematical derivations.