Computer Science student at Inha University, building AI Agents, RAG systems, and AWS serverless full-stack applications.

Currently focused on AI Agent construction and RAG/search systems while building ML fundamentals. Across 9 competition awards, I've explored diverse domains — environment, agriculture, time-series, blockchain, matching — and shipped working systems for each. Particularly interested in how retrieval, caching, indexing, and observation tie together into a single system problem in production RAG.

AI Agent · RAG FastAPI · Redis Vector Search (HNSW M=16, EF=200) · Semantic Cache · OpenAI GPT-4-turbo · AWS Bedrock · LangChain · MCP

AWS Cloud Native Lambda · API Gateway · DynamoDB · ECS Fargate · CloudFormation IaC · CodePipeline · CloudWatch

ML / Data Science Python · pandas · scikit-learn · RandomForest · XGBoost · LightGBM · pearsonr / Granger / ADF · time-series lag analysis

IM.FACT — Climate/Environment RAG Assistant Carbon Neutrality Innovation Academy 4기 (2025.05–06) 3-stage fallback RAG (Semantic Cache → Vector Search → MCP Web Search) + AWS CloudFormation IaC. Owned Frontend & Infrastructure solo, led Backend & Redis RAG engine (semantic cache + vector search). MCP module handled by another teammate.

SearchSafeJob — Safe Overseas Job Search Service DATA-UP Challenge 2nd, 2025.11 AI classifier on public APIs (외교부 · 국세청) detecting overseas job-fraud postings. Solo full-stack with ECS Fargate, FastAPI async, GitHub Actions CI/CD.

Trade Insurance Lag Effect Analysis — Big Data Competition K-SURE 1st, 2025.07 1.54M-row dataset · 12-month optimal lag discovery (162 countries statistically significant) · Ensemble of RF/XGB/LGB/SVR with R² 0.578. Caught a data-leakage issue that inflated initial R² to 0.9997 and re-derived the true generalization performance.

IGRUS BedrockGallery — First AWS Architecture Inha IGRUS Club, 2024.07–11 AWS Bedrock Gallery clone serving as my first cloud architecture design. Lambda + API Gateway + DynamoDB + Bedrock serverless.