think in systems. ship what matters.https://blog.namsang.co/kohttps://blog.namsang.co/en/posts/mcp-unnecessary-and-incomplete/https://blog.namsang.co/en/posts/mcp-unnecessary-and-incomplete/The same month MCP hit 97 million downloads, Perplexity pulled away and the Y Combinator CEO built a CLI alternative in 30 minutes. There are things the downloads do not tell you.Mon, 13 Apr 2026 15:05:00 GMTaimcpclideveloper-toolsai-agenthttps://blog.namsang.co/ja/posts/mcp-unnecessary-and-incomplete/https://blog.namsang.co/ja/posts/mcp-unnecessary-and-incomplete/9,700万ダウンロードを記録した月に、PerplexityはMCPから離脱し、Y Combinator CEOは30分でCLI代替を作った。ダウンロード数が語らないことがある。Mon, 13 Apr 2026 15:05:00 GMTaimcpclideveloper-toolsai-agenthttps://blog.namsang.co/ko/posts/mcp-unnecessary-and-incomplete/https://blog.namsang.co/ko/posts/mcp-unnecessary-and-incomplete/9,700만 다운로드를 찍은 달에, 퍼플렉시티는 이탈하고 Y콤비네이터 CEO는 30분 만에 CLI를 대안으로 만들었다. 다운로드가 말해주지 않는 것들이 있다.Mon, 13 Apr 2026 15:05:00 GMTaimcpclideveloper-toolsai-agenthttps://blog.namsang.co/en/posts/the-good-is-experience/https://blog.namsang.co/en/posts/the-good-is-experience/Dissecting failure is subtraction. Dissecting success is addition. On reproducing what worked in the next chapter.Sat, 11 Apr 2026 15:05:00 GMTengineering-leadershippost-mortemdecision-makingmental-modelsreproducibilityhttps://blog.namsang.co/ja/posts/the-good-is-experience/https://blog.namsang.co/ja/posts/the-good-is-experience/失敗を解剖するのは引き算、成功を解剖するのは足し算。次のチャプターで成功を再現することについて。Sat, 11 Apr 2026 15:05:00 GMTengineering-leadershippost-mortemdecision-makingmental-modelsreproducibilityhttps://blog.namsang.co/ko/posts/the-good-is-experience/https://blog.namsang.co/ko/posts/the-good-is-experience/실패를 해부하는 것은 뺄셈이고, 성공을 해부하는 것은 덧셈이다. 다음 챕터에서 성공을 재현하는 일에 대하여.Sat, 11 Apr 2026 15:05:00 GMTengineering-leadershippost-mortemdecision-makingmental-modelsreproducibilityhttps://blog.namsang.co/en/posts/fastapi-scoped-session-broken-pipe/https://blog.namsang.co/en/posts/fastapi-scoped-session-broken-pipe/A Broken Pipe that fell once or twice a month at dawn. pool_pre_ping=True wasn't enough. The culprit was the WSGI-era legacy baked into SQLAlchemy's scoped_session.Thu, 09 Apr 2026 15:00:00 GMTpythonfastapisqlalchemybackendhttps://blog.namsang.co/ja/posts/fastapi-scoped-session-broken-pipe/https://blog.namsang.co/ja/posts/fastapi-scoped-session-broken-pipe/月に一、二度、明け方に落ちていたBroken Pipe。pool_pre_ping=Trueでも止められなかった。犯人はSQLAlchemyのscoped_sessionに刻まれたWSGI時代のレガシーだった。Thu, 09 Apr 2026 15:00:00 GMTpythonfastapisqlalchemybackendhttps://blog.namsang.co/ko/posts/fastapi-scoped-session-broken-pipe/https://blog.namsang.co/ko/posts/fastapi-scoped-session-broken-pipe/한 달에 한두 번 새벽에 떨어지던 Broken Pipe. pool_pre_ping=True로도 막지 못했다. 범인은 SQLAlchemy의 scoped_session에 박혀 있던 WSGI 시대의 레거시였다.Thu, 09 Apr 2026 15:00:00 GMTpythonfastapisqlalchemybackendhttps://blog.namsang.co/en/posts/ship-of-theseus/https://blog.namsang.co/en/posts/ship-of-theseus/The Ship of Theseus paradox and the Strangler Fig pattern. When the last legacy code is gone, is it still the same service?Tue, 07 Apr 2026 15:00:00 GMTarchitecturestrangler-figmsalegacy-migrationrefactoringhttps://blog.namsang.co/ja/posts/ship-of-theseus/https://blog.namsang.co/ja/posts/ship-of-theseus/テセウスの船のパラドックスとStrangler Figパターン。最後のレガシーコードが消えたとき、それは同じサービスなのか?Tue, 07 Apr 2026 15:00:00 GMTarchitecturestrangler-figmsalegacy-migrationrefactoringhttps://blog.namsang.co/ko/posts/ship-of-theseus/https://blog.namsang.co/ko/posts/ship-of-theseus/테세우스의 배 역설과 Strangler Fig 패턴. 마지막 레거시 코드가 사라지면 그건 같은 서비스인가?Tue, 07 Apr 2026 15:00:00 GMTarchitecturestrangler-figmsalegacy-migrationrefactoringhttps://blog.namsang.co/en/posts/karpathy-llm-wiki/https://blog.namsang.co/en/posts/karpathy-llm-wiki/Andrej Karpathy's LLM knowledge base methodology. A raw → wiki → output 3-layer architecture, incremental compilation, and knowledge linting. The era of manipulating knowledge instead of code is arriving.Mon, 06 Apr 2026 15:00:00 GMTaiknowledge-managementobsidianllmkarpathyhttps://blog.namsang.co/ja/posts/karpathy-llm-wiki/https://blog.namsang.co/ja/posts/karpathy-llm-wiki/Andrej Karpathyが公開したLLM知識ベース方法論。raw → wiki → outputの3層、増分コンパイル、知識リンティング。コードではなく知識を操作する時代が来ている。Mon, 06 Apr 2026 15:00:00 GMTaiknowledge-managementobsidianllmkarpathyhttps://blog.namsang.co/ko/posts/karpathy-llm-wiki/https://blog.namsang.co/ko/posts/karpathy-llm-wiki/Andrej Karpathy가 공개한 LLM 지식 베이스 방법론. raw → wiki → output 3계층, 증분 컴파일, 지식 린팅. 코드가 아니라 지식을 조작하는 시대가 오고 있다.Mon, 06 Apr 2026 15:00:00 GMTaiknowledge-managementobsidianllmkarpathyhttps://blog.namsang.co/en/posts/git-worktree-ai-coding-agents/https://blog.namsang.co/en/posts/git-worktree-ai-coding-agents/git worktree is a decade-old feature, but it is gaining new relevance as infrastructure for running AI coding agents in parallel.Sat, 04 Apr 2026 15:00:00 GMTgitclaude-codeai-agentdevtoolshttps://blog.namsang.co/ja/posts/git-worktree-ai-coding-agents/https://blog.namsang.co/ja/posts/git-worktree-ai-coding-agents/git worktreeは10年前からある機能だが、Claude CodeなどのAIエージェントを並列実行するインフラとして再び脚光を浴びている。Sat, 04 Apr 2026 15:00:00 GMTgitclaude-codeai-agentdevtoolshttps://blog.namsang.co/ko/posts/git-worktree-ai-coding-agents/https://blog.namsang.co/ko/posts/git-worktree-ai-coding-agents/git worktree는 10년 된 기능이지만, Claude Code 같은 AI 에이전트를 병렬로 실행하는 인프라로 다시 주목받고 있다.Sat, 04 Apr 2026 15:00:00 GMTgitclaude-codeai-agentdevtoolshttps://blog.namsang.co/en/posts/first-90-days-review/https://blog.namsang.co/en/posts/first-90-days-review/I read it a year ago before changing jobs, and again recently to recommend to a colleague who just became a leader. Same book, different underlines.Fri, 03 Apr 2026 00:00:00 GMTbook-reviewleadershipcareeronboardinghttps://blog.namsang.co/ja/posts/first-90-days-review/https://blog.namsang.co/ja/posts/first-90-days-review/1年前、転職を控えて読んだ。最近リーダーになった同僚に薦めるために再読した。同じ本なのに、線を引く場所が変わっていた。Fri, 03 Apr 2026 00:00:00 GMTbook-reviewleadershipcareeronboardinghttps://blog.namsang.co/ko/posts/first-90-days-review/https://blog.namsang.co/ko/posts/first-90-days-review/1년 전 이직을 앞두고 읽었고, 최근 리더가 된 동료에게 추천하며 다시 읽었다. 같은 책인데 밑줄 친 곳이 달랐다.Fri, 03 Apr 2026 00:00:00 GMTbook-reviewleadershipcareeronboardinghttps://blog.namsang.co/en/posts/dena-ai-field-report/https://blog.namsang.co/en/posts/dena-ai-field-report/The numbers executives present and the reality developers experience are different. We asked a developer at DeNA directly.Thu, 02 Apr 2026 15:00:00 GMTaiengineering-cultureproductivitydenajapanhttps://blog.namsang.co/ja/posts/dena-ai-field-report/https://blog.namsang.co/ja/posts/dena-ai-field-report/経営陣が見ている数字と、現場が感じている温度は違う。DeNAで働く開発者に直接聞いた。Thu, 02 Apr 2026 15:00:00 GMTaiengineering-cultureproductivitydenajapanhttps://blog.namsang.co/ko/posts/dena-ai-field-report/https://blog.namsang.co/ko/posts/dena-ai-field-report/경영진이 보는 숫자와 현장이 느끼는 온도는 다르다. DeNA에서 일하는 개발자에게 직접 물었다.Thu, 02 Apr 2026 15:00:00 GMTaiengineering-cultureproductivitydenajapanhttps://blog.namsang.co/en/posts/dena-ai-all-in-one-year/https://blog.namsang.co/en/posts/dena-ai-all-in-one-year/In 2026, DeNA revealed the results of going "AI All-In" for one year. The scale of numbers and depth of organizational transformation suggest this is not just a tool adoption — it marks a new inflection point for the Japanese IT industry.Mon, 30 Mar 2026 16:14:00 GMTaistrategyproductivitydenajapanhttps://blog.namsang.co/ja/posts/dena-ai-all-in-one-year/https://blog.namsang.co/ja/posts/dena-ai-all-in-one-year/2026年、DeNAが"AI全振り"1年間の成果を公開した。数字の規模と組織変革の深さを見ると、これは単なるツール導入ではない。日本IT産業の新たな変曲点が始まっている。Mon, 30 Mar 2026 16:14:00 GMTaistrategyproductivitydenajapanhttps://blog.namsang.co/ko/posts/dena-ai-all-in-one-year/https://blog.namsang.co/ko/posts/dena-ai-all-in-one-year/2008년, 손정의는 스티브 잡스를 설득해 iPhone을 일본에 독점 도입했다. 일본 IT 산업의 모바일 전환은 거기서 시작됐다. 2026년, DeNA가 "AI 올인" 1년의 성과를 공개했다. 숫자의 규모와 조직 변혁의 깊이를 보면, 이건 단순한 도구 도입이 아니다.Mon, 30 Mar 2026 16:14:00 GMTaistrategyproductivitydenajapanhttps://blog.namsang.co/en/posts/rounding-is-policy/https://blog.namsang.co/en/posts/rounding-is-policy/A 1-cent discrepancy in our settlement system. Rounding is not just math — it is business policy.Fri, 27 Mar 2026 14:43:00 GMTsettlementroundingpythonkotlinhttps://blog.namsang.co/ja/posts/rounding-is-policy/https://blog.namsang.co/ja/posts/rounding-is-policy/精算システムで1円のズレが発生した。丸めは単なる算数ではなく、ビジネスポリシーである。Fri, 27 Mar 2026 14:43:00 GMTsettlementroundingpythonkotlinhttps://blog.namsang.co/ko/posts/rounding-is-policy/https://blog.namsang.co/ko/posts/rounding-is-policy/정산 시스템에서 1원 차이가 발생했다. 반올림은 단순한 수학이 아니라 비즈니스 정책이다.Fri, 27 Mar 2026 14:43:00 GMTsettlementroundingpythonkotlinhttps://blog.namsang.co/en/posts/why-individual-okrs-dont-work/https://blog.namsang.co/en/posts/why-individual-okrs-dont-work/This post is based on my experience during the OKR hype. Many organizations have since reached similar conclusions, but I wanted to document what I observed firsthand.Thu, 19 Mar 2026 15:20:00 GMTengineering-cultureokrstartupleadershiphttps://blog.namsang.co/ja/posts/why-individual-okrs-dont-work/https://blog.namsang.co/ja/posts/why-individual-okrs-dont-work/この記事は、OKRが大流行していた頃の経験をもとにしている。今では多くの組織がすでに同じ結論に達しているが、当時現場で記録した内容を振り返りながら残しておく。Thu, 19 Mar 2026 15:20:00 GMTengineering-cultureokrstartupleadershiphttps://blog.namsang.co/ko/posts/why-individual-okrs-dont-work/https://blog.namsang.co/ko/posts/why-individual-okrs-dont-work/이 글은 OKR이 한창 유행하던 시기의 경험을 바탕으로 하고 있다. 지금은 많은 조직이 이미 비슷한 결론에 도달했지만, 당시 현장에서 기록한 내용을 회고하면서 남긴다.Thu, 19 Mar 2026 15:20:00 GMTengineering-cultureokrstartupleadership