About

Hi, I'm Eliot Wong-Toi, a Senior Data Scientist at Dexcom. I earned my PhD in Statistics from the University of California, Irvine (UCI), where I was advised by Stephan Mandt.

My work sits at the intersection of statistical machine learning, uncertainty quantification, and healthcare applications. At Dexcom, I develop machine learning methods for continuous glucose monitoring (CGM) data, while my research has focused on predictive uncertainty, generative modeling, conformal prediction, and heteroskedastic regression.

Throughout my career, I have applied statistical and machine learning methods across a variety of domains, including:

• Continuous glucose monitoring (CGM): developing machine learning methods for diabetes time-series data at Dexcom, including representation learning, clustering, and glycemic pattern analysis.
• Medical imaging: applying Bayesian methods to radiomic features for skin cancer classification to support clinical decision-making.
• Clinical trial design: developing open-source software for sample size calculation and study design.
• Sports analytics: applying Bayesian nonparametric methods to analyze NBA player archetypes and offensive shot selection.
• Climate science: designing statistical tests and heteroskedastic regression methods for climate model evaluation.

Before joining Dexcom, I received my BS in Mathematics–Computer Science (cum laude) from the University of California, San Diego, where I worked in the Pulmonary Imaging Lab under Rui Carlos Pereira de Sá. There, I applied deep learning methods to medical imaging, developing tools to accelerate image annotation and segmentation. During my PhD at UC Irvine, I conducted research on uncertainty quantification in machine learning, with work spanning regression, generative modeling, and conformal prediction.

Industry

Dexcom (2025–Present)

• Develop machine learning methods for continuous glucose monitoring (CGM) data, with applications in glycemic pattern classification and user engagement modeling.
• Contribute to the development of glucotyping algorithms by improving the robustness of glycemic pattern classification across diverse longitudinal datasets.
• Designed and implemented a representation learning pipeline for clustering CGM time series from approximately 15,000 users using Vertex AI, BigQuery, and Google Cloud Storage.
• Adapted deep representation learning models for irregular, variable-length time series, improving clustering quality (silhouette score: 0.22 → 0.70).

Education

University of California, Irvine, Irvine, CA

• PhD in Statistics, Advisor: Stephan Mandt (12/2025)
• MS in Statistics (03/2022)

University of California, San Diego, La Jolla, CA

• BS in Mathematics–Computer Science (cum laude), Minor: Economics (06/2019)

Research

Dissertation

• Quantifying and Calibrating Uncertainty in Deep Predictive and Generative Models
  PhD Dissertation, University of California, Irvine, 2025
  Committee: Stephan Mandt (chair), Volodymyr Minin, Hengrui Cai

First-Author Publications

• On the Effect of Regularization on Nonparametric Mean-Variance Regression
  Eliot Wong-Toi, Alex James Boyd, Vincent Fortuin, Stephan Mandt
  (Under Review)
• Understanding Pathologies of Deep Heteroskedastic Regression
  Eliot Wong-Toi, Alex James Boyd, Vincent Fortuin, Stephan Mandt
  Uncertainty in Artificial Intelligence (UAI), 2024 (Oral)
  Earlier version presented at NeurIPS 2023 Workshop on "Machine Learning and the Physical Sciences (ML4PS)"
• A Joint Analysis for Field Goal Attempts and Percentages of Professional Basketball Players: Bayesian Nonparametric Resource.
  Eliot Wong-Toi, Hou-Cheng Yang, Weining Shen, Guanyu Hu
  Journal of Data Science, 2022

Co-Author Publications

• Generative Uncertainty in Diffusion Models
  Metod Jazbec, Eliot Wong-Toi, Guoxuan Xia, Dan Zhang, Eric Nalisnick, Stephan Mandt
  Uncertainty in Artificial Intelligence (UAI), 2025
  Best Paper Award, ICLR 2025 Workshop on "Uncertainty and Hallucination in Foundation Models"
• JANET: Joint Adaptive predictioN-region Estimation for Time-series
  Eshant English, Eliot Wong-Toi, Matteo Fontana, Stephan Mandt, Padhraic Smyth, Christoph Lippert
  Machine Learning, 2025 (ECML PKDD Journal Track)
• Navigating the Noise: Bringing Clarity to ML Parameterization Design With O(100) Ensembles
  Jerry Lin, Sungduk Yu, Liran Peng, Tom Beucler, Eliot Wong-Toi, Zeyuan Hu, Pierre Gentine, Margarita Geleta, Mike Pritchard
  Journal of Advances in Modeling Earth Systems (JAMES), 2025

Software

• MRTSampleSizeBinary (2021) — An R package for sample size calculations in micro-randomized trials.
  Eliot Wong-Toi, Thabat Dahdoul, Tianchen Qian
  CRAN package  |  Vignette  |  Shiny app

Teaching

University of California, Irvine

• Instructor of Record (Summer 2021)
  • Basic Statistics (STATS 7)
• Teaching Assistant (2019–2021)
  • Basic Statistics (STATS 7)
  • Introduction to Biological Statistics (STATS 8)
  • Introduction to Probability & Statistics for Computer Science (STATS 67)
  • Probability & Statistics I (STATS 120A)
• Reader (2020)
  • Probability & Statistics II (STATS 120B)

University of California, San Diego

• Supplemental Instruction Leader (2017–2018)
  • Precalculus (MATH 3A)
  • Calculus-Based Introductory Probability & Statistics (MATH 11)
  • Linear Algebra (MATH 18)
  • Vector Calculus (MATH 20E)

Service

• Committee Member, What Matters to Me and Why Speakers Series (UCI, 2022–2025)
• Mentor, LEAP Summer REU Program and Momentum Fellowship (LEAP, 2024)
• Mentor, Diverse Educational Community and Doctoral Experience 1st Year Program (UCI, 2022–2024)
• Panelist, Graduate School Informational Panel (UCI, 2023)
• Panelist, Data Science REU Graduate School Panel (UCI, 2023)
• Panelist, Teaching Assistant Professional Development Program Panel (UCI, 2021)
• Reviewer for NeurIPS, ICML, ICLR, AISTATS, TMLR