CV

Education

• Ph.D. Candidate in Applied Mathematics, University of Washington, 2021 - 2026 (expected)
  • Advisors: Prof. Lillian Ratliff (ECE), Prof. Eric Shea-Brown (AMATH)
  • GPA: 4.0
  • Boeing Research Award (2024)
• B.Sc. in Mathematics and Computer Science, University of Wisconsin-Madison, 2017 - 2021
  • GPA: 4.0
  • Dean’s Prize Nominee (top 51 students in the class of 2021)
  • ICPC World Finalist (2019)
  • MCM Meritorious Winner - Top 10% (2018)

Industry Experience

• Student Researcher Intern, Ai2, Seattle, WA (Nov 2025 - Present)
  • Curated and standardized 1,500+ heterogeneous LeRobot robotics datasets into a unified corpus (19M frames, 40k episodes)
  • Auto-annotated robot instructions with Qwen-2.5VL vision-language model
  • Proposed cold-start pretraining strategy: reduced pretraining loss by 50%, improved pick-up success rate by 67%
• Research Scientist Intern, Meta, New York, NY (Jun 2024 - Dec 2024)
  • Scaled neural interface foundation model to 1B+ parameters, improving handwriting success rate by 39%
  • Learned discretized EMG representations via Gumbel-Softmax
  • Implemented global-local attention masking with Torch FlexAttention (40× throughput improvement)
• Enterprise Software Engineer Intern, Facebook Inc., Menlo Park, CA (May 2019 - Aug 2019)
  • Developed web interface for purchase order management using React, JavaScript, Relay, and Hack
  • Contributed 3000+ lines of code, received return offer

Research Interests

Generative Models, Neural Network Generalization, Reinforcement Learning, Computational Neuroscience, Machine Learning

Selected Publications

1. Shirui Chen, Jiantao Jiao, Lillian J. Ratliff, Banghua Zhu. “dUltra: Ultra-Fast Diffusion Language Models via Reinforcement Learning”

2. Shirui Chen, Stefano Recanatesi, Eric Shea-Brown. “A simple connection from loss flatness to compressed representations in neural networks”. TMLR 2025

3. Shirui Chen, Linxing Jiang, P.N. Rajesh Rao, Eric Shea-Brown. “Expressive probabilistic sampling in recurrent neural networks”. NeurIPS 2023

4. Linxing Preston Jiang, Shirui Chen, et al. “Data Heterogeneity Limits the Scaling Effect of Pretraining Neural Data Transformers”. COLM 2025, LM4Sci Workshop

5. Shirui Chen, Qixin Yang, Sukbin Lim. “Efficient inference of synaptic plasticity rule with Gaussian process regression”. iScience 2023

Awards

• Boeing Research Award (2024)
• Violet Higgitt Frank Scholarship (2020)
• ICPC World Finalist (2019)
• MCM Meritorious Winner - Top 10% out of 10,670 teams (2018)

Service

• Reviewer for NeurIPS, ICLR, ICML
• Co-reviewer for Current Opinions In Neurobiology

Technical Skills

• Programming & Tools: Python, C++, Go, MATLAB, NumPy, Pandas, Linux, Git
• Frameworks: PyTorch, Accelerate, TRL, Torch FlexAttention
• Expertise: Deep Learning, Generative Models, Reinforcement Learning, Computational Neuroscience, Bayesian Inference & Statistics