Usan - Build and Publish Workspace

┌──────────────────────────────────┐
│   CHEOTJUL · 첫줄                │
│   ─────────────────────────────  │
│   The first line of your build.  │
│   cha_oo / since 2024-09         │
└──────────────────────────────────┘

cheotjul.sh — bash

session: 12.4k commits · 187 builds · 2024-09 → 2026-05

> uptime618d 12:04

> last buildgpt-4o-mini · 12s ago

> queue3 pending

> regionseoul · gmt+9

$ now() ≈ alive

> ./services

Four AI automations. No quote needed — first 1-hour diagnostic is free.

> Quotes scale with company size and data volume · talk it through at #contact

> ./build_pipeline

Diagnose → design → PoC → run. Four stages, 4-6 weeks on average.

[step 01]

┌──────┐
│  01  │
└──────┘

Diagnose

1 wk

Map current stack, problems, KPIs. Two technical meetings (video). Output: one diagnostic report.

[step 02]

┌──────┐
│  02  │
└──────┘

Design

1 wk

Three solution options + quote. Proceed after your decision. Output: SOW + schedule.

[step 03]

┌──────┐
│  03  │
└──────┘

PoC

2-4 wk

Code, deploy, demo, hand-off doc. Output: code repo + demo link + run guide.

[step 04]

┌──────┐
│  04  │
└──────┘

Run

ongoing

Monthly maintenance, tuning, monitoring. Or in-house transfer (1-week training free).

> All code is your asset · NDA optional · no lock-in rule

> ./about/cha_oo

Ex-fintech backend, 7 years → went solo on AI automation in Sep 2024.

In 2018 I joined a fintech startup as a backend engineer. For seven years I built payment, settlement and KYC systems — a daily loop of SQL queries, python scripts and Slack automation.

In spring 2024 I built my first in-house AI automation with LangChain. One customer-response task that took 30 minutes ran in one minute with an LLM, and more accurately. That settled it — more Korean companies should be building this.

This is a one-person company. Not a large SI shop — it helps you build a PoC fast and decide fast. The 1-hour diagnostic is free.

All code is your asset. No lock-in rule. Run it together, or transfer it in-house — both work.

> cha_oo / Cheotjul founder / since 2024-09

> ./stack

4 primary + 6 secondary. Best mix tuned to your environment.

// primary

python

langchain · langgraph

last_used: today

typescript

cursor · claude

last_used: today

postgres

pgvector

last_used: 2d ago

n8n

make · zapier

last_used: today

// secondary

rust

performance critical

last_used: 6d ago

high-throughput

last_used: 3d ago

dagster

data orchestration

last_used: 4d ago

chromadb

vector store alt

last_used: 2d ago

supabase

auth + db PoC

last_used: today

deploy

vercel · fly · render

last_used: today

> ./faq

The five things first-time clients ask most.

KRW 120,000/hour is mid-tier for the Korean market. The 1-hour diagnostic is free, so compare first, then decide. The point is no lock-in and a full code hand-off, not the rate.

Every PoC runs in your cloud (AWS / GCP / your own infra). No Cheotjul server. No data sent out. NDA signed first on request. Enterprise-plan vendor options are also walked through.

Yes. Code and docs are all handed off, with a 1-week training included (no extra fee). No Cheotjul lock-in rule — it is designed for you to run it at your own pace.

Multiple LLM options, given as a first recommendation + comparison table tuned to domain, budget and data-residency needs — organized by Korean quality, cost and data residency.

Progress is shared weekly (git log + monitoring dashboard access). One person, but two backup collaborators on informal standby — a 4-hour response SLA on urgent outages, which can be faster than a large SI.

> ./contact

Request the free 1-hour diagnostic. Reply within 24 hours.

> ./services

1-Hour Diagnostic

LLM Chatbot · RAG

n8n · Make Automation

MCP Server Build

> ./build_pipeline

> ./about/cha_oo

> ./stack

> ./faq

> ./contact