DataNexus Build Journal

DataNexus Series more

9. The Public Benchmark Returned 56%: Nine Experiments and What Got Ruled Out
8. From 66% to 80% NL2SQL Accuracy: Four Measure-and-Fix Loops
7. When a Question Comes In, Who Decides the Routing?
6. When You Don't Have to Build Agent Infra Yourself, Harnesses Become Obsolete. What About the Ontology?
5. Automating Metadata Maintenance: Karpathy's LLM Wiki Architecture

DW Modeling more

4. Super-Sub Types — Can a Customer Be Both Individual and Corporate?
3. ERD Notation — Same Diagram, Different Interpretation
2. OLTP vs DW Models — Different Purpose, Different Design
1. Is Kimball Still Relevant in the Cloud DW Era?

ETL Design more

4. SCD - When a Customer Moves, Where Were Past Orders Shipped?
3. Silver Layer - Promoting Bronze to an Analysis-Ready State
2. Bronze Layer - Load the Source Data Exactly As-Is
1. Medallion Architecture - Why We Stack Data in Three Layers

GEO Optimization more

5. AEO - Why Coding Agents Read Documentation Differently
4. Off-Site GEO - How to Win Over AI That Ignores Your Official Site
3. On-Site GEO Technical Architecture - From Product DB to JSON-LD
2. Each AI Cites Different Sources
1. What Is GEO - AI Citation Strategy Beyond SEO

Curations more

What Actually Improves Claude Code Performance: Configuration and Architecture
32 Slash Command Shortcuts That LLMs Instantly Understand
Analyzing the Official GitHub Repository of a Major Korean Brokerage Open API
macshot: A Native macOS Tool Emerging as an Alternative to Paid Apps
OpenDocuments: A Local RAG Platform That Unifies Fragmented Team Knowledge

- views

- visitors

GoatCounter GA4 Dashboard

Roadmap
Posts
Guides
Curations
Tags
Categories
About

Aperag

2. How We Chose These 4 Open-Source Tools

How we decided on DataHub + Vanna + ApeRAG + DozerDB for DataNexus. What got eliminated from the candidate list, and why.

February 17, 2026 · 7 min · Junho Lee

« Prev Next »

© 2026 Junho Lee · Powered by Hugo & PaperMod