Jieun

Deservewhatyouwant.

All Articles

Systems 2026-10-18

Different Topics, So Why Are We Getting the Same Sources?

I traced similar sources across different topics back to the actual search queries and code. Grok expresses the question, while code preserves search intent and assigns platform sources.
Systems 2026-10-18

How Can We Make AI-Generated Blog Titles Readable?

Letting Grok judge meaning while code preserves and validates titles
Systems 2026-09-23

Redefining the Product Direction After Beta Feedback

How beta users' questions and behavior changed Wiblog's customer hypothesis, onboarding, B2C/B2B direction, and value proposition.
Systems 2026-09-18

From Topic Idea to Generated, Published, Search-Visible Content

From topic ideas to publishing and search visibility
Systems 2026-09-15

Why We Moved Development from Four 16GB MacBooks to One 96GB Server

Separate workspaces and permissions, with one shared API tunnel
Systems 2026-09-08

The In-App Browser Eats Your Social Login Button — and Hard Logins Lose Users Instantly

Detect, escape, fall back — for drop-offs no log catches
DevOps 2026-09-08

Switching LLM Accounts at 88% — Manually First, Then Automatic Failover at 80%

From a Grafana 88% alert to automatic failover at 80%
Reliability 2026-08-25

The Domain Was Purchased but Would Not Load — Automating Connections Around DNS Delays

Separating immediate setup from work that waits for DNS, with connection checks and registration retries
DevOps 2026-08-18

How I Cut AI Blog Generation Costs by Reducing Duplicate Searches and Failed Retries

Recomputing 15 topics from logs: ₩309 down to ₩150
Systems 2026-08-17

Eight Cheap Calls Beat One Good One — Why I Removed the Model That Wrote Better

Why fast repeated Grok calls beat the better writer
Systems 2026-08-10

Published but Missing from Search — Why I Started Recommending Custom Domains

How indexing observations changed Wiblog's domain recommendations
Systems 2026-08-10

Keyword Discovery: 25 Seconds Down to 0.2, by Calling Search Autocomplete

Replacing two LLM calls with one autocomplete request
Reliability 2026-07-20

Taking Caddy Out of the Deploy Path — the Structural Problem Was Touching a SPOF Automatically

Removing 60–100s outages, then the last 5 seconds
Reliability 2026-05-17

[Messenger Load-Testing, Part 3] FCM to a Queue, Swallowed Failures to Sentry — Cleaning Up the Message-Send Handler

Isolating external I/O and surfacing swallowed errors
Reliability 2026-05-11

[Messenger Load-Testing, Part 2] There Was a Deadlock Hiding in AI-Generated Code — How We Replaced `unreadCount` with `lastReadAt`

42.8% errors in 9 minutes, caused by one updateMany
Scaling 2026-05-04

[Messenger Load-Testing, Part 1] Building a Reusable Load-Test Environment — Make-based IaC, Observability, and Auth Seeding

A 7-node production clone, rebuilt with one make command
Reliability 2026-03-29

Why You Should Never Use KeyDB as a Pub/Sub Broker for Centrifugo

Five years of hang issues behind a perfect-looking active-replica
Reliability 2026-03-26

Building a Crash-Resilient, End-to-End Meeting Recording Pipeline in the Browser

Surviving a 300MB heap and tab crashes with 5-minute savepoints