Resume

Edison Xu

Software Engineer & AI/ML Engineer

edisonapply@gmail.com626-552-1696Atlanta, GA, USALinkedIn GitHub

Summary

I'm Edison, a software engineer building AI-powered products at the intersection of full-stack engineering, retrieval, and agent systems. At Highmark Health, I've worked on real-time AI training platforms, planner–executor–critic multi-agent research workflows, and document intelligence tools on top of Vertex AI Search. My focus is designing systems where LLMs, retrieval, and infrastructure work together to produce grounded, auditable, and reliable behavior in production.

Experience

Software Engineer (Generative AI team) · Highmark Health

Feb 2025 - Present

Built production-grade internal AI systems spanning real-time interaction, multi-agent research, and enterprise retrieval workflows. Shipped a real-time call-center training platform, a citation-backed Planner–Executor–Critic deep research system, and an enterprise document Q&A workspace using grounded retrieval, reranking, and Vertex AI Search—while contributing to CI/CD, Cloud Run deployments, and observability for reliability at scale.

Real-time AIMulti-Agent SystemsRetrieval / RAGVertex AI SearchCloud RunCI/CD

Software Engineer Intern (Generative AI team) · Highmark Health

May 2024 - Dec 2024

Contributed to core AI platform and retrieval capabilities across internal tools, model integrations, and source-grounded workflows. Built a multi-modal platform on Vertex AI, a RAG-based policy assistant for grounded summarization, and a centralized Gemini proxy service—while improving reliability with Pytest coverage above 90%.

Vertex AIRetrieval / RAGGemini APIAI PlatformPolicy Q&ATesting

Education

Master of Science in Computational Science & Engineering

Georgia Institute of Technology

2022-2024 · December 2024

Bachelor of Arts in Applied Economics

University of California - Santa Barbara

2019-2021 · March 2021

Skills

AI SystemsLLMs, Multi-Agent Systems, LangGraph, Vertex AI, Gemini API, RAG Systems, Embedding Models

Web & ProductReact.js, Next.js, Vue.js, Spring Boot, Django, Node.js, Express.js, Selenium, FastAPI, Flask

LanguagesPython, Go, TypeScript, JavaScript, Java, SQL

Cloud & DataMySQL, PostgreSQL, MongoDB, Redis, DynamoDB, Firebase, Docker, Kubernetes, Git, AWS, GCP, Cloud Run, GitOps, CI/CD, Postman, Nginx, Linux, Jira, PySpark

Selected Projects

FigBrain: Figma Multi-Agent Plugin

Built a Figma AI agent at the Microsoft AI Hackathon, enabling natural-language prompts to generate and populate FigJam components via the Model Context Protocol (MCP) and automated Figma API/plugin flows. Designed a supervisor multi-agent system using LangGraph, where a master orchestrator routed tasks to planner, executor, and verifier agents with Chain-of-Thought prompts and ReAct loops, validated at ~85% first-try accuracy. Cut API costs ~28% and lowered latency through long/short-term memory mechanism, plan/result caching, and dynamic model switching between lightweight and advanced LLMs.

Golang-based High-Efficiency Search Engine

Developed a Go-based search engine with text and vector search capabilities, using gRPC for microservices architecture. Implemented a distributed web crawler with Kafka cluster integration for efficient data collection and processing. Enhanced search relevance using TF-IDF and BM25 ranking algorithms, with Trie trees for optimized query suggestions.

CloudBin: Cloud-Based Distributed File Storage System

Designed a storage system using Spring Boot, Vue3 and AWS S3, featuring chunk-based file uploads for reliability. Leveraged Redis with a Bloom Filter to prevent cache penetration, boosting cache hit rate and reducing database load. Integrated ElasticSearch for efficient fuzzy file name searches, reducing search time and improving accuracy.