KO / EN / JA

About Me

Introduction

An AI Engineer who has designed and deployed services using diverse AI technologies including Vision, Audio, and LLM. Gained broad experience from service deployment to PoC across domains such as children's English learning, metaverse, and B2B solutions.

Built comprehensive experience across LLM, image, speech, and recommender systems — from model training and inference optimization to deployment. Capable of end-to-end development including from-scratch training pipelines and cloud deployment with Azure ML.

Also pursues maximizing development productivity with AI tools. Regularly uses Claude Code, GitHub Copilot, and more. This portfolio site itself was designed and implemented through collaboration with Claude Code. Expert not only in building AI, but also in working with AI.

LLM / Agent Computer Vision Speech (STT/TTS) Search System Recommender MCP Unreal Engine MLOps PyTorch Azure ML FastAPI

Experience

Career

2026.02 ~ Present

Wemade Next

AI EngineerMIR5 Programming Office, Server Team

MIR5

  • [Unreal MCP] Built an MCP server enabling natural language CRUD operations on objects in Unreal Engine, designing a scalable architecture with async processing, N:M connection structure, and dynamic tool registration
  • [MCP Client] Developed a lightweight client supporting chat and MCP integration using subscription-based models (Codex, GitHub Copilot, etc.) without API keys

2024.04 ~ 2025.06

Creverse (Chungdahm Learning)

AI EngineerNesto CV Division, AI Biz Team

AI Auto-Assessment System

  • [MLOps] Designed and deployed a real-time inference pipeline using Azure ML Endpoint and Blob Storage, operating an auto-assessment service used by ~30K students monthly
  • [LLM Agent] Implemented a synthetic student data generation pipeline using LangGraph-based multi-agent architecture
  • [Audio] Fine-tuned NVIDIA Parakeet STT on non-native children's speech data, reducing WER from 9.6% to 4.4%
  • [Vision] Developed a presentation attitude recognition model using Google MediaPipe face/pose tracking, achieving 98.32% accuracy and 98.77% F1
  • [Vision] Built a Diffusion model-based image generation pipeline with quantization for model lightweighting
  • [Search] Implemented a hybrid search system combining keyword BM25 + vector search, verified ~20% reduction in LLM API calls

2022.10 ~ 2024.01

Metaverse World (Netmarble F&C)

AI Research EngineerDev Division 1, MetaWorld Dev Part

Grand Cross: MetaWorld

  • [Vision] Designed and implemented a Face-to-Character prototype for generating game avatars from user face images by analyzing and referencing research papers (PoC)
  • [Vision] Developed motion tracking by applying MediaPipe-based Pose Estimation to reflect user's real-time movements on 3D characters

2021.06 ~ 2022.04

Coconut Silo

AI Research EngineerCTO Office

  • [Recommender System] Embedded product descriptions using BERT and implemented a content-based filtering (CBF) recommendation system using inner product similarity
  • [Vision] Fine-tuned YOLOv5 for ship and obstacle detection in maritime radar imagery
  • [Anomaly Detection] Designed and implemented a vision-based anomaly detection system for real-time analysis of metal 3D printer process footage

2020.08 ~ 2021.05

Gaion Co., Ltd.

Data ScientistService Business Division

  • [Data] Data collection and processing for government-funded project (Data Voucher) analytics
  • [Planning] Planning and data collection/processing/analysis for an SME matching platform using trade data

Skills

Tech Stack

🧠

Frameworks & Libraries

PyTorchPyTorch LightningHuggingFaceTransformersLangChainLangGraphFastAPIGradioStreamlitW&B
💻

Languages

PythonC++
☁️

Cloud & Infra

Azure MLBlob StorageDockerWSL
📊

Data & Vision

OpenCVFFmpegPandasPolarsNumPyMatplotlibSeaborn
🛠️

Dev Tools

Visual StudioVS CodeGitHubPerforceJiraConfluenceSlack
🤖

AI-Powered Dev

Claude CodeGitHub Copilot
🎮

Others

Unreal Engine

Education

Education

Sogang University, AI·SW Graduate School

Data Science & AI, M.Eng.

2023.03 ~ 2026.02

Academic Excellence Scholarship (Fall 2023)

Thesis (2026.01)

줄거리 임베딩 결합을 통한 LightGCN 기반 영화 추천 시스템의 성능 향상 연구

Improving LightGCN-based Movie Recommendation Performance via Plot Embedding Integration

Advisor: Jeongmin So

Recommender Systems GNN LightGCN LLM
View Original

Sejong University

Physics, B.S.

Certifications

AICE (AI Certificate for Everyone), Basic 2022.11
ADsP (Advanced Data Analytics Semi-Professional) 2022.06

AI-Augmented Development

Curious about AI-augmented development?

From design to development to testing — an introduction to AI-powered dev workflows.

15+

Projects

36

Pipeline Steps

4+

Years Exp.

How I Work with AI