Career Profile
I currently work as a Data Scientist / AI Engineer at Samsung SDS, focused on LLM application development, including RAG systems and agentic workflows. Currently employed, and open to new opportunities.
Before joining Samsung SDS, I completed my Ph.D. in Mathematical Sciences at KAIST, where I studied deep neural networks with inter-node dependency and heavy-tailed distributions. I have a solid foundation in probability theory and statistics, which I apply to develop robust AI solutions.
Professional interests:
- Large language models and their applications, including retrieval-augmented generation (RAG) and AI agents
- Data analysis, machine learning, and deep neural networks with heavy-tailed behavior
- Probability theory – infinite and finite exchangeability, random graphs and graph limits
Knowledge & Skills:
- Theoretical foundations – Probability Theory, Deep Learning, Statistics
- Core skills – LangChain & LangGraph, FastAPI, PyTorch
Experiences
Build and deploy LLM-powered applications for enterprise use cases. Design RAG pipelines and agentic workflows with strong emphasis on evaluation, reliability, and safety. Use DevOps tools for CI/CD, deployment, and monitoring in production environments.
- Developed RAG systems with data ingestion, retrieval, reranking, and prompt orchestration
- Implemented agent tools and workflows for multi-step reasoning and task automation
- Built a no-code data analysis tool using a code interpreter for automated exploratory analysis and reporting
- Established evaluation frameworks for LLM quality, groundedness, and latency
- Delivered APIs and services using FastAPI, integrating with cloud platforms including Azure and Samsung Cloud Platform
Education
- Thesis: Infinitely wide neural networks with heavy tails and inter-node dependence
- Advisor: Prof. Paul Jung
Focused on the theory of deep neural networks with heavy tails and inter-node dependence. Studied limiting behaviors of network initialization and learning dynamics, and their connections to kernel methods.
Keywords: Neural Network Gaussian Process (NNGP), Neural Tangent Kernel (NTK).
Theoretical background in probabilistic tools including stochastic processes, exchangeable structures, concentration inequalities, and random graphs and graph limits.