maxu@illinois.edu

Google Scholar

Github

Linkedin

Curriculum Vitae (CV)

I am a final year PhD student that is actively searching for full-time positions in industry and academia. Please contact me via my @illinois.edu email on the left sidebar.

I work at the intersection of applied and theoretical machine learning for time-series data, with a strong application focus on wearable health biosensors. My goal is to develop self-supervised learning methodologies to be used with foundation models for mobile health.

As a University of Illinois Urbana-Champaign PhD candidate, I work with Jim Rehg as my advisor and have recieved the NSF Graduate Research Fellowship and Georgia Tech Presidential Fellowship awards. I started my PhD with Dr. Rehg at Georgia Tech before transferring to UIUC with him.

I am currently working on Google Health Research as a student researcher, researching foundation models for wearable health biosignals for their FitBit devices. Prior to this, I interned at Apple Health AI where I spearhead the research and de-risking of their motion foundation model, which has now been implemented on device for the AirPod Pro 3's new calorie and activity tracking capabilities.

In Review at ICLR 2026

LSM-2: Learning from Incomplete Wearable Sensor Data

LSM-2 introduces Adaptive and Inherited Masking, which learns from incomplete data without imputation. Trained on 40M hours, it outperforms in classification, regression, and generation while robustly handling real-world missingness.

Paper

ICLR 2025

RelCon: Relative Contrastive Learning for a Motion Foundation Model for Wearable Data

RelCon is a self-supervised motion foundation model for IMU sensor data, achieving state-of-the-art performance across tasks like activity recognition and gait analysis. It’s the first generalizable model of its kind, trained on 1B segments from 87K participants.

Paper Code Poster Press

ICLR 2024

REBAR: Retrieval-based Reconstruction for Time-series Contrastive Learning

We introduce a novel method of identifying positives/negatives for time-series contrastive learning with REtrieval-BAsed Reconstruction. REBAR predicts mutual class membership via motif similarities, achieving SOTA performance.

Paper Code Poster

NeurIPS 2022

PulseImpute: A Novel Benchmark Task for Pulsative Physiological Signal Imputation

PulseImpute is the first mHealth pulsative signal imputation challenge which includes realistic missingness models, clinical downstream tasks, and an extensive set of baselines, including an augmented transformer that achieves SOTA performance.

Paper Project Page Code

Aug 2020 — Ph.D. in Computer Science
PresentUniversity of Illinois Urbana-Champaign
Advisor: Jim Rehg
Anticipated Graduation: Spring 2026

Aug 2020 — M.S. in Computer Science
May 2023Georgia Institute of Technology
Overall GPA: 4.00/4.00

Aug 2016 — B.S. in Applied Mathematics and Statistics
Dec 2019 B.S. in Biomedical Engineering
Johns Hopkins University
Overall GPA: 3.84/4.00

2024Apple AIML PhD Fellowship Finalist
• One of 3 candidates nominated by UIUC for invitation-only fellowship submission.
• Nominated as an emerging leader for AI for Health and Wellness.

20242nd Place Poster at the 11th Annual Jump ARCHES Symposium
• Awarded for poster on RelCon: Relative Contrastive Learning for a Motion Foundation Model for Wearable Data.
• Jump ARCHES is a partnership between OSF Healthcare and UIUC that partners teams of clinicians and engineers working together to improve patient outcomes and reduce health care costs.

2022National Science Foundation Graduate Research Fellowship
• Awarded for displaying exemplary levels of research skill and potential
• Press Release: https://grad.gatech.edu/news/49-students-awarded-prestigious-national-science-foundation-graduate-research-fellowship

2020Georgia Tech Presidential Fellowship
• Awarded for displaying exemplary levels of scholarship and innovation

2020Intuitive Surgical Best Project Award
• For "MATHIAS: Modeling A Transferable Histopathological Image Analysis System"
• Press Release: https://www.cs.jhu.edu/2020/01/28/deep-learning-course-prepares-students-for-success-in-ai-careers/

20191st Place in Fall 2019 Johns Hopkins FastForward U Spark Accelerator Competition
• Press Release: https://ventures.jhu.edu/news/bioswift-aquatas-fast-forward-u-accelerator-demo-days/

20191st Place in 2019 ASAIOfyi Student Design Competition
• Press Release: https://www.facebook.com/jhu.admissions/posts/congrats-to-team-bioswift-jhubme-hopkinsengineering/10156494724603946/

LSM-2: Learning from Incomplete Wearable Sensor Data
Maxwell A Xu^†, Girish Narayanswamy^†, Kumar Ayush, Dimitris Spathis, Shun Liao, Shyam A Tailor, Ahmed Metwally, A Ali Heydari, Yuwei Zhang, Jake Garrison, Samy Abdel-Ghaffar, Xuhai Xu, Ken Gu, Jacob Sunshine, Ming-Zher Poh, Yun Liu, Tim Althoff, Shrikanth Narayanan, Pushmeet Kohli, Mark Malhotra, Shwetak Patel, Yuzhe Yang, James M Rehg, Xin Liu°, Daniel McDuff°
In Review at ICLR, 2026
Paper ^†Co-first authors °Co-last authors

OpenTSLM: Time-Series Language Models for Reasoning over Multivariate Medical Text- and Time-Series Data
Patrick Langer, Thomas Kaar, Max Rosenblattl, Maxwell A Xu, Winnie Chow, ... , Daniel McDuff, Elgar Fleisch, Oliver Aalami, Filipe Barata, Paul Schmiedmayer
In Review at ICLR, 2026
Paper Code Press

The Anatomy of a Personal Health Agent
A. Ali Heydari^†, Ken Gu^†, Vidya Srinivas^†, Hong Yu^†, ... , Maxwell A Xu, ... , Xin Liu°, Daniel McDuff°, Xuhai "Orson" Xu°
In Review at Nature, 2026
Paper ^†Co-first authors °Co-last authors

Leveraging Shared Prototypes for a Multimodal Pulse Motion Foundation Model
Wanting Mao^†, Maxwell A Xu^†, Harish Haresamudram, Mithun Saha, Santosh Kumar, James Matthew Rehg
In Review at ICLR, 2026
Paper ^†Co-first authors

Pulse-PPG: An Open-Source Field-Trained PPG Foundation Model for Wearable Applications Across Lab and Field Settings
Mithun Saha^†, Maxwell A Xu^†, Wanting Mao, Sameer Neupane, James M Rehg, Santosh Kumar
Ubiquitous Computing (UbiComp), Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), 2025
Paper ^†Co-first authors

SensorLM: Learning the Language of Wearable Sensors
Yuwei Zhang, Kumar Ayush, Siyuan Qiao, A Ali Heydari, Girish Narayanswamy, Maxwell A Xu, Ahmed A Metwally, Shawn Xu, Jake Garrison, Xuhai Xu, Tim Althoff, Yun Liu, Pushmeet Kohli, Jiening Zhan, Mark Malhotra, Shwetak Patel, Cecilia Mascolo, Xin Liu, Daniel McDuff, Yuzhe Yang
Proceedings of the Neural Information Processing Systems (NeurIPS) 2025
Paper

RelCon: Relative Contrastive Learning for a Motion Foundation Model for Wearable Data
Maxwell A. Xu, Jaya Narain, Gregory Darnell, Haraldur Hallgrimsson, Hyewon Jeong, Darren Forde, Richard Fineman, Karthik J Raghuram, James M Rehg, Shirley Ren
The Thirteenth International Conference on Learning Representations (ICLR), 2025
Paper Code Poster Press

Towards Time Series Reasoning with LLMs
Winnie Chow, Lauren Gardiner, Haraldur T Hallgrímsson, Maxwell A. Xu, Shirley You Ren
Oral Presentation at NeurIPS Workshop on Time Series in the Age of Large Models, 2024
Paper

Temporally Multi-Scale Sparse Self-Attention for Physical Activity Data Imputation
Hui Wei, Maxwell A. Xu, Colin Samplawski, Hui Wei, James M. Rehg, Santosh Kumar, Benjamin M. Marlin
Conference on Health, Inference, and Learning (CHIL), 2024
Paper

REBAR: Retrieval-Based Reconstruction For Time-series Contrastive Learning
Maxwell A. Xu, Alexander Moreno, Hui Wei, Benjamin M. Marlin, James M. Rehg
The Twelfth International Conference on Learning Representations (ICLR), 2024
Paper Code Poster

PulseImpute: A Novel Benchmark Task for Pulsative Physiological Signal Imputation
Maxwell A. Xu, Alexander Moreno, Supriya Nagesh, V. Burak Aydemir, David W. Wetter, Santosh Kumar, James M. Rehg
Proceedings of the Neural Information Processing Systems (NeurIPS), 2022.
Paper Project Page Code Poster

Discovering Novel Predictors of Minimally Verbal Outcomes in Autism through Computational Modeling
Maxwell A. Xu, James M. Rehg, Agata Rozga, Jena McDaniel, Paul Yoder, Linda R. Watson, Nancy Brady
INSAR Oral + Press Conference (~ 1% acceptance rate), 2022.
Paper Press Release Video

Efficient Learning and Decoding of the Continuous-Time Hidden Markov Model for Disease Progression Modeling
Yu-Ying Liu, Alexander Moreno^†, Maxwell A. Xu^†, Jena McDaniel, Nancy Brady, Agata Rozga, Fuxin Li, Le Song, James M. Rehg
arXiv, 2021.
Paper Code ^†Co-second authors

Impact of Sequencing Radiation Therapy and Immune Checkpoint Inhibitors in the Treatment of Melanoma Brain Metastases
Daniel Pomeranz Krummel^‡, Tahseen H. Nasti^‡, Benjamin Izar^†, Robert H. Press^†, Maxwell Xu^†, Lindsey Lowder, Laura Kallay, Manali Rupji, Havi Rosen, Jing Su, Walter Curran, Jeffrey Olson, Brent Weinberg, Matthew Schniederjan, Stewart Neill, David Lawson, Jeanne Kowalski, Mohammed Khan, Soma Sengupta
International Journal of Radiation Oncology, 2020.
Paper ^‡Co-first authors; ^†Co-second authors

Modulating Native GABAA Receptors in Medulloblastoma with Positive Allosteric Benzodiazepine-derivatives Induces Cell Death
Laura Kallay, Havva Keskin, Alexandra Ross, Manali Rupji, Olivia A Moody, Xin Wang, Guanguan Li, Taukir Ahmed, Farjana Rashid, Michael Rajesh Stephen, Kirsten A Cottrill, T Austin Nuckols, Maxwell Xu, Deborah E Martinson, Frank Tranghese, Yanxin Pei, James M Cook, Jeanne Kowalski, Michael D Taylor, Andrew Jenkins, Daniel A Pomeranz Krummel, Soma Sengupta
Journal of Neuro-Oncology, 2019.
Paper

Dec 2024 —Google, Seattle, WA
PresentStudent Researcher, Google Health
• Research on foundation models for wearable health biosignals

Apr 2024 —Apple, Seattle, WA
Sep 2024Research Intern, Health AI
• Spearheaded research efforts on self-supervised learning methods for a health time-series foundation model.
• Collaborated with interdisciplinary team to fuse health-domain knowledge in guiding model development
• Heavily assisted in engineering efforts for machine learning pipeline and infrastructure

Jan 2020 —Systems & Technology Research, Boston, MA
Aug 2020Machine Learning Researcher, Cyber-Physical Systems
• Spearheaded machine learning initiatives within the cybersecurity vulnerability research
• Created a reinforcement learning method for greybox mutation-based fuzzer
• Developed a seq2seq VQ-VAE WaveRNN decoder for unsupervised representation learning of audio

May 2019 —Medtronic, North Haven, CT
Dec 2019AI/Data Science Engineer Contractor, Minimal Invasive Therapies Group
• Developed a tool to predict lung cancer recurrence from clinical big data using machine learning
• Created a computer vision blob detection tool
• Utilized survival analysis with Kaplan-Meier curve visualizations to identify cancer recurrence risk factors

Aug 2018 —BioSwift, Baltimore, MD
Dec 2019Former Chief Executive Officer and Co-Founder
• Designed device to augment dry powder inhalers for usage among pediatric asthmatics
• Secured over $10,000 in funding, from various business and design competitions as well as the Johns Hopkins Student Initiatives Fund
• Press Release: https://www.jhunewsletter.com/article/2019/11/fastforward-u-teams-innovate-with-new-and-old-technologies

May 2018 —Centers for Disease Control and Prevention, Atlanta, GA
Aug 2018Biostatistics Intern, Outbreak Surveillance
• Spearheaded a new initiative to utilize whole genome multilocus sequence typing data for detecting disease clusters

EXTH-12. Radiation Enhances Melanoma Response to Immunotherapy and Synergizes with Benzodiazepines to Promote Anti-Tumor Activity
Daniel Pomeranz Krummel^‡, Tahseen H. Nasti^‡, Benjamin Izar^†, Robert H. Press^†, Maxwell Xu^†, Lindsey Lowder, Laura Kallay, Manali Rupji, Havi Rosen, Jing Su, Walter Curran, Jeffrey Olson, Brent Weinberg, Matthew Schniederjan, Stewart Neill, David Lawson, Jeanne Kowalski, Mohammed Khan, Soma Sengupta
Neuro-Oncology 2020. Phoenix, Arizona
Paper ^‡Co-first authors; ^†Co-second authors

Abstract 247: Identification of the GABAA Receptor in Melanoma Brain Metastases Patient Tumors and Demonstration that It is a Viable Drug Target using Benzodiazepine-Derivatives
Milota Kaluzova, Tahseen Nasti, Hiao-Rong Chen, Lindsey Lowder, Robert Press, Havi Rosen, Manali Rupji, Laura Kallay, Rikesh Patel, Andre Burnham, Maxwell Xu, Alexandra Ross, Havva Keskin, Erin Connelly, Benjamin Izar, Cory Adamson, Jeffrey Olson, Jing Su, Walter Curran, Ragini Kudchadkar, Matthew Schniederjan, Stewart Neill, David Lawson, Michael Chan, Jeanne Kowalski, Mohammad Khan, Daniel Pomeranz Krummel, Soma Sengupta
AACR Annual Meeting 2019. Atlanta, GA
Paper

Abstract 2623: Modulating Native GABAA Receptors in Medulloblastoma with Positive Allosteric Benzodiazepine-Derivatives Induces Cell Death
Laura Kallay, Havva Keskin, Alexandra Ross, Manali Rupji, Olivia A. Moody, Xin Wang, Guanguan Li, Taukir Ahmed, Farjana Rashid, Michael Rajesh Stephen, Kirsten A. Cottrill, Austin Nuckois, Maxwell Xu, Deborah E. Martinson, Frank Tranghese, Yanxin Pei, James M. Cook, Jeanne Kowalski, Michael D. Taylor, Andrew Jenkins, Daniel Pomeranz Krummel, Soma Sengupta
AACR Annual Meeting 2019. Atlanta, GA
Paper

COMP-22: Large Scale Transcriptomic Analysis of Melanoma Brain Metastases
Jeanne Kowalski, Daniel Pomeranz Krummel, Manali Rupji, Bhakti Dwivedi, Havva Keskin, Laura Kallay, Maxwell Xu, Alexandra Ross, Robert Press, Havi Rosen, Erin Connelly, Rikesh Patel, Benjamin Izar, Cory Adamson, Jeffrey Olson, Jing Su, Ragini Kudchadkar, Matthew Schniederjan, Lindsey Lowder, Stewart Neill, Walter Curran, David Lawson, Michael Chan, Mohammad Khan, Soma Sengupta
Neuro-Oncology 2018. New Orleans, LA
Paper

PDTM-45: Positive Modulation of Native GABAA Receptors in Medulloblastoma Cancer Cells with Benzodiazepines Induces Rapid Mitochondrial Fragmentation and TP53-Dependent, Cell Cycle-Independent Apoptosis
Laura Kallay, Havva Keskin, Alexandra Ross, Olivia Moody, Kirsten Cottrill, Austin Nuckols, Guanguan Li, Taukir Ahmed, Farjana Rashid, Michael Stephen, Maxwell Xu, Deborah Martinson, Tobey Macdonald, Jeanne Kowalski, Xin Wang, Michael Taylor, James Cook, Andrew Jenkins, Daniel Pomeranz Krummel, Soma Sengupta
Neuro-Oncology 2018. New Orleans, LA
Paper

CD131: Large Scale Transcriptomic Analysis of Melanoma Brain Metastases
Jeanne Kowalski, Daniel Pomeranz Krummel, Manali Rupji, Bhakti Dwivedi, Havva Keskin, Laura Kallay, Maxwell Xu, Alexandra Ross, Robert Press, Havi Rosen, Erin Connelly, Rikesh Patel, Benjamin Izar, Cory Adamson, Jeffrey Olson, Jing Su, Ragini Kudchadkar, Matthew Schniederjan, Lindsey Lowder, Stewart Neill, Walter Curran, David Lawson, Michael Chan, Mohammad Khan, Soma Sengupta
Annals of Neurology 2018.
Paper