# Xinye Yang (Charlie)

> Software Development Engineer at AWS Kiro and research assistant at the University of Colorado Anschutz School of Medicine. Research on AI for healthcare: EHR-based patient world models, reliable medical imaging AI, clinical AI agents, and the data infrastructure to train them. Engineering work on LLM context and memory systems for agentic coding tools.

Xinye Yang (Chinese name: 杨新烨) also publishes as "X Yang" and "Xin-Ye Yang", and is known as Charlie Yang. He received an M.S. in Computer Science from Brown University (2026) and a B.Eng. in Artificial Intelligence from the University of Science and Technology Beijing (2024).

## Profiles

- [Homepage](https://yangxinyee.github.io/) · [中文主页](https://yangxinyee.github.io/zh/)
- [Google Scholar](https://scholar.google.com/citations?user=xEgmvx4AAAAJ)
- [ORCID 0009-0004-3143-4263](https://orcid.org/0009-0004-3143-4263)
- [Semantic Scholar](https://www.semanticscholar.org/author/2370872191)
- [GitHub](https://github.com/Yangxinyee)
- [LinkedIn](https://www.linkedin.com/in/xinye-charlie-yang-939674286/)
- [CV (PDF)](https://yangxinyee.github.io/assets/Yang_Xinye_CV.pdf)

## Research interests

Multimodal LLMs; medical image analysis; AI in healthcare; agentic AI; EHR-based patient world models; reliable and parameter-efficient AI for clinical decision support and hospital triage.

## Open-source projects

- [EHR2Trace](https://yangxinyee.github.io/projects/ehr2trace/) ([code](https://github.com/Yangxinyee/ehr2trace)): converts hospital EHR exports into OMOP CDM 5.4 and MEDS from one canonical layer, with row-level source lineage, separate event and availability times, and 55 validation checks on the published outputs. Built for patient world models and clinical agents. Apache-2.0. The project page compares it with MEDS-Extract, meds_etl, ehr2meds and the OHDSI MIMIC-IV ETL.
- [vrm-edge-triage](https://github.com/Yangxinyee/vrm-edge-triage): code for the Smart Health 2026 paper on confidence-gated cloud-edge cascade triage via variational risk minimization (VRM) for chest X-ray triage.
- [cxr-vlm-routing](https://github.com/Yangxinyee/cxr-vlm-routing): code, MIMIC-CXR per-case results and tables for the CHASE 2026 workshop paper on reliability stress tests and decision-time routing for chest X-ray vision-language models (CheXagent, MedGemma-4B, MedGemma-27B). Apache-2.0.
- [Q-DISTILL](https://github.com/Yangxinyee/Q-DISTILL): self-supervised Q-Former distillation for image-only chest X-ray triage using MedGemma pseudo-reports; accuracy 76.5% to 89.1%. Part of the Smart Health 2026 work.
- [brain_mri_preprocess_pipeline](https://github.com/Yangxinyee/brain_mri_preprocess_pipeline): stroke brain MRI preprocessing (DICOM to NIfTI, registration, skull stripping, encryption).

## Blog

- [Stress-Testing Chest X-ray Vision-Language Models Before Deployment](https://yangxinyee.github.io/blog/cxr-vlm-stress-test/) (2026-09-28): how to test CheXagent- and MedGemma-style chest X-ray VLMs across prompts, workflows and metrics, why exact match misleads, when multi-agent pipelines help or hurt, and per-case routing. Based on the CHASE 2026 workshop paper and the cxr-vlm-routing repository.
- [How to Keep Data Lineage When Building EHR Training Datasets](https://yangxinyee.github.io/blog/ehr-data-lineage/) (2026-09-28): eight practices for auditable EHR training data for patient world models and clinical agents, with examples from EHR2Trace.

## Publications

- Xinye Yang, Zhusi Zhong, Scott Collins, Michael Bernstein, Grayson Baird, Terrence Healey, Michael Atalay, Mahesh Jayaraman, Xuyu Wang, Zhicheng Jiao. Confidence-gated cloud-edge cascade triage via variational risk minimization for medical imaging. Smart Health 41, 100689 (2026). Presented as an oral at IEEE/ACM CHASE 2026. https://doi.org/10.1016/j.smhl.2026.100689
- Xinye Yang, Zhusi Zhong, Scott Collins, Grayson Baird, Xuyu Wang, Zhicheng Jiao. Reliability Stress Tests and Decision-Time Routing for Chest X-ray Vision-Language Models. IEEE/ACM CHASE 2026 Workshop. https://doi.org/10.1109/CHASE69719.2026.00073 Code: https://github.com/Yangxinyee/cxr-vlm-routing
- Zhuoqi Ma, Xinye Yang, et al. The AI Challenge: A Turing Test Pilot Study of Attendings and Residents in Identifying AI-Generated Content. Meta-Radiology (2025). https://doi.org/10.1016/j.metrad.2025.100199
- Zhuoqi Ma, Xinye Yang, et al. A Unified Platform for Radiology Report Generation and Clinician-Centered AI Evaluation. medRxiv (2025). https://doi.org/10.1101/2025.07.07.25331018
- Cheng Chen, Huilin Wang, Yunqing Chen, Zihan Yin, Xinye Yang, et al. Understanding the brain with attention: A survey of transformers in brain sciences. Brain-X (2023). https://doi.org/10.1002/brx2.29
- Xinye Yang, Yuhang Liu, Zhiwei Lin, Lingyu Zhong, Teoh Teik Toe. Optimization of CNN for Diagnosis on Lung Disease by Lung Segmentation and Rib Suppression. ISCID 2022. https://doi.org/10.1109/iscid56505.2022.00063

## Experience

- 2026-present: Software Development Engineer, Amazon Web Services (Kiro), Seattle.
- 2026-present: Research Assistant, University of Colorado Anschutz School of Medicine (advised by Prof. Harrison Bai and Prof. Yuli Wang).
- 2025: Software Engineer Intern, AWS Kiro; Research Intern, Brown University Health.
- 2023-2024: Software & ML Engineer Intern, Institute of Automation, Chinese Academy of Sciences.