I'm an AI Engineer focused on RAG systems, LLM applications, retrieval evaluation, and backend deployment.

Focus Areas: Hybrid Retrieval (BM25 + Dense), RAG Evaluation (RAGAS), Vector DB (ChromaDB), Sentence Transformers

Korean financial-domain RAG chatbot using Bank of Korea glossary documents.

• Hybrid Retrieval & LLM: Implemented sparse/dense hybrid retrieval and Ollama-based QA pipeline.
• Full-Stack & Auth: Built FastAPI backend and React dashboard with JWT/RBAC authentication.
• Monitoring & Eval: Integrated stage-level latency/throughput monitoring and RAGAS-based evaluation.
• Deployment: Containerized applications using Docker and successfully deployed via GHCR to Render/Vercel.

Transformer-based English-Korean translation model implemented from scratch.

• Scratch Implementation: Built modular Encoder, Decoder, and Attention components using PyTorch.
• Pipeline: Developed data preprocessing, training, evaluation, and inference entry points.
• Objective: Focused on deep understanding of model internals without relying on pretrained APIs.

• RAG & LLM Application Development
• Search, Retrieval, and Ranking Systems
• AI Backend Infrastructure & Model Evaluation