About me

AI/ML software engineer with experience in both core and applied AI research, now building production AI systems from the ground up.

I'm Joohan Lee, a Senior AI Application Developer at Samsung Electronics America with a blend of research experience and practical industry expertise. I specialize in AI applications, RAG pipelines, full-stack development, and AI research to create innovative solutions.

Key Achievements

  • Enterprise LLM Platform - Built and run the company-wide LLM platform at Samsung Electronics America, scaled from a 36-user pilot to 1,600+ monthly and 800+ daily active users
  • 1st Place Winner at IEEE ICASSP Signal Processing Grand Challenges (ML competition)
  • Open-Source Contributor - Merged a billing fix into LibreChat (45k+ stars); also contributed to LibreCodeInterpreter and the Open WebUI docs
  • Enterprise AI Impact - Led enterprise-level development of an AI-based mortgage assistant application
  • IEEE Published Researcher with 4 peer-reviewed papers including top-tier IEEE TWC journal publication
  • USC Computer Science Master's graduate with focus on AI/ML and Deep Learning

What Sets Me Apart

  • Research-to-Production Pipeline: Successfully transitioned from academic research at USC to enterprise AI applications, bringing theoretical ML/AI knowledge into real-world scalable solutions
  • End-to-End AI Development: Complete experience from ML model training (PyTorch, TensorFlow) to production deployment (AWS, Docker) and user interfaces (React, Next.js)
  • Turning Ideas into Reality: Strong background in computer science and AI, along with hands-on backend and full-stack development experience
  • Quick Learner: Loves diving into new challenges and is always eager to explore innovative tools and ideas

Technical Expertise

  • AI/ML: LLM Platform Engineering, AI Agents, AWS Bedrock, RAG, LangChain, HuggingFace, PyTorch, TensorFlow, Milvus, Deep Learning, NLP
  • Programming: Python, TypeScript/JavaScript, C/C++, Java, SQL, HTML/CSS
  • Backend & Cloud: FastAPI, Flask, Node.js, RESTful APIs, MongoDB, MySQL, AWS (EC2, Lambda, Bedrock), Linux, Docker, Nginx, Gunicorn
  • Frontend: React, Next.js, Electron, HTML/CSS

What I'm doing

  • LLM Platform Engineering

    Building and running a company-wide LLM platform: a self-hosted LibreChat fork and a FastAPI gateway on AWS Bedrock Claude, with SSO and audit logging.

  • AI Agents & Automation

    Multi-agent systems (ReAct, orchestrator, human-in-the-loop) and a workflow orchestration platform that turns business processes into AI agents, with sandboxed execution and data isolation.

  • AI Research & Applied ML

    Channel prediction, language models for physical layer communications, and noise-robust ML, published in IEEE venues.

  • Full-Stack Development

    End-to-end AI products with React, Next.js, FastAPI, Node.js, and MongoDB on AWS.

Resume

Education

  1. University of Southern California

    Jan 2022 — Dec 2023
    Master of Science in Computer Science
  2. Yonsei University

    Mar 2015 — Feb 2021
    Bachelor of Engineering in Computer Engineering

Professional Experience

  1. Senior AI Application Developer

    Aug 2026 — Present
    Samsung Electronics America, Plano, TX

    Building and running company-wide AI platforms: a self-hosted LLM chat platform on AWS Bedrock and a workflow orchestration platform that turns business processes into AI agents and automation.

    Key Contributions:
    • Company-wide LLM Platform: Built and run a self-hosted LibreChat fork and a FastAPI inference gateway on AWS Bedrock Claude, scaled from a 36-user pilot to 1,600+ monthly and 800+ daily active users and 1.1M+ monthly inference calls
    • Workflow Orchestration Platform: Building a platform that turns 21 processes (760 tasks) of a 100+ person business operations organization into AI agents and automation
    • Secure AI Agents: Engineered AI agents with sandboxed execution tools, data isolation, and private-cloud data residency
    Technologies: LibreChat, FastAPI, AWS Bedrock, Claude, Python, AI Agents, Sandboxed Execution
  2. Gen AI Software Engineer

    Oct 2024 — Aug 2026
    Samsung SDS America, Ridgefield Park, NJ

    Led development of enterprise-scale generative AI applications, multimodal RAG pipelines, and LLM-powered analytics systems. Specializing in cutting-edge AI technologies for business solutions.

    Key Contributions:
    • Full-stack AI Integration: Built AI applications with authentication, chatbot, document processing, and user management
    • Multimodal RAG Pipeline: Designed and implemented retrieval-augmented generation enhancing domain-specific information retrieval and generation performance
    • Performance Improvement: Achieved 20% improvement in response accuracy through advanced optimization techniques
    • Security Enhancement: Identified and resolved 15+ security vulnerabilities in production AI application
    • Cloud Deployment: Deployed production applications using AWS EC2 with EBS, Application Load Balancer, Route53, S3, VPC, CloudWatch, Nginx, and Gunicorn
    • Data Analytics: Developed a sales data analyzer enabling users to analyze sales data with natural language using LLM, Pandas, and LangChain, incorporating prompt engineering and agentic tools
    • Multi-Agent Analytics Platform: Drove architecture and development of a multi-agent data analytics platform built on ReAct, orchestrator, and human-in-the-loop (HITL) patterns, which converts user questions into visualization, insights, and business strategy, saving thousands of hours of manual analysis per year
    Technologies: LangChain, HuggingFace, Milvus, AWS Bedrock, PyTorch, TensorFlow, Flask, React, Electron
  3. Full-Stack Developer

    Apr 2024 — Sep 2024
    USP Management, Inc, Fort Lee, NJ

    Developed comprehensive ERP systems for logistics data management, implementing both backend APIs and frontend interfaces for streamlined business operations.

    Key Contributions:
    • RESTful API Development: Designed and implemented APIs improving operational excellence for logistics data management
    • ERP UI Implementation: Created user-friendly interfaces resulting in streamlined user interactions and improved data accessibility
    • Full-Stack Architecture: Built end-to-end solutions using modern web technologies
    Technologies: Node.js, Express.js, TypeScript, MySQL, Next.js, React
  4. Graduate Research Assistant

    Dec 2022 — May 2024
    University of Southern California, Los Angeles, CA

    Conducted cutting-edge research in AI communications and machine learning, developing novel approaches for robust communication systems and noise-robust learning frameworks.

    Research Contributions:
    • LLM Communication Framework: Built on-device AI communication framework using TensorFlow to simulate physical layer communications
    • Vector Quantization: Applied VQ techniques achieving 10% noise-robustness improvement and 50% compression
    • ML Pipeline Development: Created diverse ML pipelines using PyTorch and TensorFlow improving model accuracy by 18%
    • Publications: Multiple IEEE papers published in prestigious venues
    Technologies: PyTorch, TensorFlow, Python, NumPy, Linux, Slurm, HPC
  5. Software Engineer Intern

    May 2023 — Jul 2023
    CyClean, Inc, Los Angeles, CA (Remote)

    Developed RESTful APIs and dynamic web applications for IoT products, gaining hands-on experience with cloud deployment and modern web technologies.

    Key Contributions:
    • API Development: Implemented RESTful APIs for customer information and quotes about IoT products using AWS Lambda
    • Web Application: Developed dynamic website utilizing React and TypeScript
    • Cloud Deployment: Deployed applications using AWS Amplify, CloudFront, and Route53
    Technologies: JavaScript, React, TypeScript, AWS Lambda, AWS Amplify, CloudFront, Route53
  6. Research Intern

    May 2022 — Apr 2023
    Purdue University, Los Angeles, CA (Remote)

    Contributed to software entity recognition research, developing WikiSER dataset and noise-robust learning approaches for NLP applications.

    Research Contributions:
    • WikiSER Dataset: Developed high-quality dataset of 1.7M sentences with 79K labeled software entities
    • Noise-Robust Learning: Implemented self-regularization approach achieving 7.1% F1 score improvement
    • Error Reduction: Achieved 8.62% reduction in error rate compared to existing benchmarks
    Technologies: Python, NLP, Machine Learning, Data Processing, Research Methodologies
  7. Software Engineer

    Jul 2020 — Jul 2021
    BSG One, Seoul, South Korea

    Developed JSON standard interfaces and led multiple projects, gaining valuable experience in software development and cross-functional team collaboration.

    Key Contributions:
    • Interface Development: Developed JSON standard interface for BrityRPA using JavaScript
    • Productivity Improvement: Increased development productivity by up to 30%
    • Project Leadership: Led 6 projects end-to-end with excellent client communication
    Technologies: JavaScript, JSON APIs, RPA Development, Project Management

Open-Source Contributions

Technical Skills

AI/ML & LLM Engineering

LLM Platform Engineering, AI Agents (ReAct, multi-agent, HITL), AWS Bedrock, RAG, Prompt Engineering, LangChain, HuggingFace, PyTorch, TensorFlow, NLP, Deep Learning, Computer Vision

Programming Languages

Python, TypeScript/JavaScript, C/C++, Java, SQL

Backend & APIs

FastAPI, Flask, Node.js, Express.js, RESTful APIs, Nginx, Gunicorn, Linux

Security & Auth

ADFS/OIDC SSO, Audit Logging, Sandboxed Code Execution, Data Isolation

Frontend

React, Next.js, Electron, HTML/CSS

Cloud & DevOps

AWS (EC2, S3, VPC, Lambda, Route53, CloudFront, Bedrock), Docker, Git

Databases

MongoDB, MySQL, Milvus (Vector DB)

Networking & Wireless

TCP/IP, OSPF, BGP, Socket Programming, Congestion Control (BBR), 5G NR Link-level Simulation (Sionna), MIMO Channel Prediction, Physical Layer Communications, On-device AI

Other Domains

Enterprise AI Platforms

Portfolio

Publications