Ziyuan Guan
Data Scientist III · Systems Architect

Ziyuan Guan

IC3 & PRISMAp Lab · College of Medicine, University of Florida · Gainesville, FL

Building data infrastructure and systems for clinical AI — ETL pipelines, EHR data wrangling, and embedding models into production. Leading data and systems cores across multiple NIH-funded projects.

hub GitHub work LinkedIn school Scholar mail Contact
33+
Publications
300+
Citations
6+
NIH Grants
7+
Years Exp.
Open Source

Featured Projects

View All
terminal ClinicClaw
Rust

AI agents for real clinical workflows — ambient documentation, smart orders, and prior authorization automation.

security VERITAS
Rust

Deny-by-default policy engine with SHA-256 audit trail for AI agent governance in clinical environments.

schema HAVEN
TypeSpec

Health Asset Value & Exchange Network — patient data sovereignty protocol with consent, provenance, and value tracking.

science PSDL
Python

Patient Scenario Definition Language — deterministic, portable language for clinical decision support and research scenarios.

Career

Experience

Full Timeline
Data Scientist III — IC3, College of Medicine, UF
Nov 2022 – Present

Lead Data Systems Architect across multiple NIH-funded initiatives (Bridge2AI-ChoRUS, MEnD-AKI, Intelligent-ICU, XAI-IDEALIST). Manages cloud resources (Azure, AWS), HPC allocations, and data pipelines for 14M+ encounter datasets.

PythonKafkaSparkFHIR R4Kubernetes
System Admin & Data Analyst — IC3, UF
Sep 2021 – Nov 2022

Distributed Linux systems with UF Health IT. EHR ETL pipelines with Python, Spark, Kafka. Built React + Flutter web/mobile apps for MySurgeryRisk.

ReactFlutterMongoDB
System & Data Analyst — PRISMAp Lab, UF
Sep 2019 – Sep 2021

MySurgeryRisk platform — PHP/JS frontend, Java RESTful APIs, Spark/Kafka AI engine. 1000% caching improvement via Memcache/Redis.

JavaSparkCassandra
Research

Selected Publications

All 33 Pubs
2025
MANDARIN: Mixture-of-Experts for Dynamic Delirium and Coma Prediction in the ICU
Contreras et al. · arXiv:2503.06059
2025
Federated Learning for Predicting Major Postoperative Complications
Ren et al. · Annals of Surgery Open, 6(2), e573
2024
Risk-specific training cohorts to address class imbalance in surgical risk prediction
Balch et al. · JAMA Surgery, 159(12), 1424–1431
2024
Electronic Health Record Data Quality and Performance Assessments
Penev et al. · JMIR Medical Informatics, 12(1), e58130
2022
Performance of a ML algorithm using EHR data to predict postoperative complications
Ren et al. · JAMA Network Open, 5(5), e2211973
Skills

Tech Stack

Full Stack
Languages
DartPythonRustJavaTypeScriptSQLC++
Frameworks
ReactFlutterDjangoSpringBootFHIR R4
Data & Cloud
KafkaSparkAirflowAWSAzureHiPerGator
Databases
PostgreSQLMongoDBCassandraMySQL
Academic

Education

school
MS Electrical & Computer Engineering
University of Florida · Jul 2017 – May 2019
school
BS Electrical & Computer Engineering
New York Institute of Technology · Jul 2013 – May 2017
school
BS Electrical & Computer Engineering
Nanjing University of Posts and Telecommunications · Jul 2013 – May 2017
mail