DeokJung Kim eyelove

LLM Wiki

A pattern for building personal knowledge bases using LLMs.

This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, OpenCode / Pi, or etc.). Its goal is to communicate the high level idea, but your agent will build out the specifics in collaboration with you.

The core idea

Most people's experience with LLMs and documents looks like RAG: you upload a collection of files, the LLM retrieves relevant chunks at query time, and generates an answer. This works, but the LLM is rediscovering knowledge from scratch on every question. There's no accumulation. Ask a subtle question that requires synthesizing five documents, and the LLM has to find and piece together the relevant fragments every time. Nothing is built up. NotebookLM, ChatGPT file uploads, and most RAG systems work this way.

AI Coding Agent Guidelines (claude.md)

These rules define how an AI coding agent should plan, execute, verify, communicate, and recover when working in a real codebase. Optimize for correctness, minimalism, and developer experience.

Operating Principles (Non-Negotiable)

Correctness over cleverness: Prefer boring, readable solutions that are easy to maintain.
Smallest change that works: Minimize blast radius; don't refactor adjacent code unless it meaningfully reduces risk or complexity.

NAVER DEVIEW 2017 Slides

키노트 영상

DAY 1

Requirements

Minikube requires that VT-x/AMD-v virtualization is enabled in BIOS. To check that this is enabled on OSX / macOS run:

sysctl -a | grep machdep.cpu.features | grep VMX

If there's output, you're good!

Prerequisites

한글과 유니코드

유니코드에서 한글을 어떻게 다루는지를 정리하였다.

유니코드

유니코드(Unicode)는 전 세계의 모든 문자를 컴퓨터에서 일관되게 표현하고 다룰 수 있도록 설계된 산업 표준 (위키 백과)
단순히 문자마다 번호를 붙임
계속 업데이트되며 현재는 Unicode Version 9.0.0 이 최신이다.

UTF

유니코드를 실제 파일 등에 어떻게 기록할 것인지를 표준화한 것이다.

	[
	// =======================================================
	// 본 자료는 CC0-1.0 라이선스를 따릅니다.
	// https://creativecommons.org/publicdomain/zero/1.0/deed
	// 저작권자의 허락을 구하지 않아도 상업적인 목적을 포함한 어떠한 목적으로든
	// 자유롭게 복사, 수정, 배포, 실연할 수 있습니다.
	// =======================================================
	// 모든 정보는 직접 수집하여 정리한 정보입니다.
	// 일부 정보가 오기되었거나, 위/경도 좌표가 부정확할 수 있습니다.
	// Gist 댓글을 통해 오류를 제보해주세요.

	from pyspark.sql.functions import col, explode_outer
	from pyspark.sql.types import *
	from copy import deepcopy
	from autoflatten import AutoFlatten
	from collections import Counter

	s3_path = 's3://mybucket/orders/'
	df = spark.read.orc(s3_path)
	json_df = spark.read.json(df.rdd.map(lambda row: row.json))
	json_schema = json_df.schema

	import json

	class AutoFlatten:
	def __init__(self, json_schema):
	self.fields_in_json = self.get_fields_in_json(json_schema)
	self.all_fields = {}
	self.cols_to_explode = set()
	self.structure = {}
	self.order = []
	self.bottom_to_top = {}

	##### READ SPARK DATAFRAME
	df = spark.read.option("header", "true").option("inferSchema", "true").csv(fname)
	# store the schema from the CSV w/ the header in the first file, and infer the types for the columns
	df_schema = df.schema

	##### SAVE JSON SCHEMA INTO S3 / BLOB STORAGE
	# save the schema to load from the streaming job, which we will load during the next job
	dbutils.fs.rm("/home/mwc/airline_schema.json", True)

	with open("/dbfs/home/mwc/airline_schema.json", "w") as f:

	#!/bin/bash
	com=${1:-"htop"}
	tname=mon
	twin=mon
	tnw=$tname:$twin

	tmux has-session -t $tname
	if [ $? != 0 ]
	then
	tmux new-session -s $tname -n $twin -d ssh -t a1 "$com"