James Cui

James Cui

Software Engineer · AI Engineer

I build production software and agentic systems — currently working on healthcare infrastructure at Epic Systems, and exploring multi-agent reasoning, RAG/MCP pipelines, and using AI to automate complex workflows.

Email: jamescui226@gmail.com Phone: 978-505-6459 Location: Verona, WI
Now
Software Developer @ Epic Systems
Building healthcare software for 500+ hospital networks.
Focus
Multi-Agent Systems · RAG · MCP
Agentic pipelines, tooling, reliability.
Background
UW–Madison CS + Math
GPA 3.92 · Graduated Aug 2025

Experience

Software Developer

Epic Systems · Verona, WI

Jan 2026 – Present

  • Engineered an opt-in filing model and pop-up interface to prevent surgical history clutter for hundreds of surgeons; grouped redundant procedures and enabled discretionary filing to conserve thousands of postoperative hours across 15+ hospitals.
  • Cut data entry latency for 500+ healthcare networks by architecting a React, C#, and TypeScript interface; reconciled legacy data to assist nurses in populating missing vendor fields, streamlining supply-chain workflows and eliminating supply waste.

Software Engineer (AI/ML)

PaperScore · Austin, TX

Sept 2025 – Dec 2025

  • Architected a multi-agentic evaluation pipeline utilizing RAG and MCP to autonomously determine manuscript novelty.
  • Scaled data ingestion 30× by developing an agentic web scraper (Browser Use) capable of indexing 30+ journals per hour.
  • Boosted reliability from 25% to 80% by refactoring the C# backend with semaphores and rate-limiting to manage API traffic.

Computer Vision Research Assistant

MadAbility Research Group · Madison, WI

Jun 2024 – May 2025

  • Architected a real-time computer vision pipeline utilizing YOLOv8 for sub-second object detection (faucets, sinks) to trigger restrictive visual augmentations for in-site OCD self-management and compulsion disengagement.
  • Outperformed COCO accuracy by 15% by curating a dataset of 1,250+ images and fine-tuning a YOLOv8 model in PyTorch.
  • Improved clinical intervention success by 60% by designing an AI-driven AR system in Unity/C# that integrated Exposure and Response Prevention (ERP) and Acceptance and Commitment Therapy strategies into real-time patient environments.

Technical Skills

Languages

Python Java TypeScript C# C/C++ PostgreSQL NoSQL Go Bash

AI & Agents

MCP RAG Browser Use LangChain LLM Orchestration PyTorch TensorFlow Transformers NLP LSTM

Frameworks & Tools

Spring Boot React Next.js Node.js FastMCP Playwright Docker Kubernetes AWS (EC2/S3/Lambda) Git Claude Code CI/CD Linux/Unix JUnit

Projects

News Trader

Verona, WI

Built a high-throughput Python Asyncio engine processing 5k+ daily signals using task queues. Engineered a multi-agentic "council" using various LLMs to autonomously classify geopolitical and corporate events — achieved a 160% return on paper-traded capital over a 1-month validation period during the war in Iran.

Python Asyncio Multi-Agent LLMs pytest
View Code →

Smart Stock

Madison, WI

Full-stack Java/Spring Boot inventory system with a RESTful API handling concurrent inventory updates. Hardened security via Spring Security role-based access control and containerized application services with Docker.

Java Spring Boot Docker React SQL

Outlier

Scaled a full-stack word puzzle platform to 100+ active users with a responsive React/TypeScript frontend. Eliminated 100% of manual content operations by architecting an autonomous pipeline using Gemini API and GitHub Actions — generating and deploying daily puzzles of varying difficulty, reducing operational overhead by 20+ hours/week.

TypeScript React Python Firebase Auth Gemini API GitHub Actions

Neural Machine Translator

Hong Kong S.A.R.

Achieved a 4× improvement in BLEU scores by fine-tuning a pre-trained BART transformer using Hugging Face. Accelerated model training cycles by 40% by engineering a high-concurrency PyTorch pipeline with custom tokenization.

Python PyTorch Transformers NLP NumPy
View Code →

Education

University of Wisconsin–Madison

B.S. Computer Science · Certificate in Mathematics · Completed in 3 years

Aug 2025

GPA: 3.92 / 4.0 Academic Excellence Scholar Dean's List (x5)
Relevant coursework
Algorithms Software Engineering Operating Systems Data Structures Artificial Intelligence Deep Learning Databases Computer Networks OOP Computer Graphics Combinatorics

Beyond Code

A few things I care about outside of work.

🏐

Volleyball

Volleyball has been a big part of my life — I play competitively and love the team dynamics, quick decision-making, and the grind of improving every rep. Whether it's pickup games or organized leagues, you can usually find me on the court on weekends.

Competitive Team Player Always Improving
🤖

Currently Learning

In my free time I'm deep-diving into multi-agent thinking — how to design systems where multiple AI agents reason, debate, and collaborate to solve problems no single model can. I'm also experimenting with using Claude to control my computer, building real workflows where Claude writes code, runs tools, and ships things end-to-end with minimal human input.

Multi-Agent Systems Claude Code Computer Use Agentic Workflows
▶️

YouTube

I put project demos and build videos on YouTube. Check out my Smart Stock inventory system demo — a full walkthrough of a production-grade Spring Boot + React app with Docker and Spring Security.

Watch Smart Stock Demo →
📚

Currently Reading

  • The Mom Test Rob Fitzpatrick — how to talk to customers and validate ideas
  • Zero to One Peter Thiel — building companies that create new things
  • Designing Machine Learning Systems Chip Huyen — ML system design in production

Always open to recommendations.

✈️

Places I Want to Go

Bucket list destinations — mixing culture, food, and new experiences.

🇯🇵
Japan
Tokyo · Kyoto · Osaka
🇮🇹
Italy
Rome · Amalfi Coast
🇧🇷
Brazil
Rio de Janeiro
🇹🇭
Thailand
Bangkok · Chiang Mai
🇵🇹
Portugal
Lisbon · Porto
🇦🇷
Argentina
Buenos Aires · Patagonia