I start with the shape of the problem and the limits of the data.
Kunho Pro
Designing intelligent systems.
I explore how LLMs, RAG, and AI agents can find, understand, and act on complex information with reliable structure.
Designed for
LLM, RAG, Agent
I connect papers, experiments, and reproducible pipelines quickly.
I care about whether model behavior holds up in real user flows.
Core architecture
Deep reasoning. Clear output.
I study data and machine learning at Pusan National University, with a focus on retrieval-augmented generation, LLM applications, and automated reasoning pipelines.
RAG
LLM
Monitoring
AI Agent
Built-in apps
Featured projects.
Financial RAG System
A reliable RAG pipeline for reading and reasoning over financial PDFs.
ColPali - BM25 - BGE-M3 - Gemini Vision View detailsCustomizing LoRA
Style adaptation experiments for Stable Diffusion with LoRA.
LoRA - Stable Diffusion - Diffusers - PEFT View detailsRetail Demand Forecasting
Demand forecasting and inventory optimization at store and SKU level.
TFT - XGBoost - Safety Stock View detailsTelco Churn Prediction
Churn prediction with monitoring from prototype to retraining.
XGBoost - Evidently AI - Retraining View detailsPersonal mode
Curiosity is the default spec.
I document what I learn, test assumptions until they become clear, and prefer systems with a simple structure beneath the interface.
Technical specifications
Kunho specs.
- Model
- Jang Kunho - AI Researcher
- Born
- 2003.12.18
- Physical Spec
- 192cm
- Core stack
- Python, PyTorch, TensorFlow, Scikit-learn
- Interested in
- LLM, RAG, AI Agent, Machine Learning Systems