Hi, I'm
Sr. Staff Engineer & Bar Raiser @ Coupang
When Execution becomes cheap, Judgment is the moat — but Judgment without Ground Truth is just guessing
Learn MoreI define the What and deliver the How. Built infrastructure from zero, open-sourced tools with 5000+ Stars, shipped production AI systems — across three technology cycles: systems engineering, mobile, and AI.
Cross-domain thinking is how I work: understanding LLMs through the lens of compilers, interpreting AI consciousness through cognitive science, making technical decisions with a contrarian investor's logic. 30+ deep-dive articles on AI, one core thesis — when Execution becomes cheap, Judgment is the real moat.
"The ceiling of How depends on the precision of What."
— Johnson
Shipping production AI systems like Autonomous AB Clean at Coupang, while producing 30+ deep-dive articles. What I care about is not how many lines of code AI can write, but how Judgment becomes the core competitive advantage when Execution becomes cheap.
Production AI practice at Coupang — an autonomous AB experiment cleanup system that embeds AI Agents into the development workflow, automating experiment lifecycle management
From Agent architecture to hands-on Claude Code practice, exploring optimal human-AI collaboration. Proposed "Harness Engineering" — building order from chaos
Using probabilistic tools to verify probabilistic outputs equals no verification. The most underrated AI-era advantage: deterministic validation systems
LLMs are functions, not databases. From compilers to LLMs — understanding the cyclical nature of software abstraction layers
When Agents become the entry point, attention economy pivots to ROI economy. Apps retreat to backend infrastructure; software engineering faces its final paradigm revolution
Those who can decompose vague requirements into precise specifications are the scarcest talent in the AI era. The ceiling of How depends on the precision of What
Built company-wide mobile infra from scratch, launched R.LUX luxury app from scratch, empowered engineering with AI. Cross-team architecture decisions, root cause analysis of performance bottlenecks, tech debt prioritization; as Bar Raiser, making independent hiring judgments beyond the hiring manager.
"Defining technical direction, upholding talent standards — ensuring every new hire raises the bar."
Large-scale mobile architecture design and performance optimization.
"With hundreds of millions of DAU in short-video, there are no shortcuts to peak experience optimization."
Led driver/rider mobile platform architecture and open-sourced the Booster build optimization tool (5000+ GitHub Stars).
"Solid infrastructure isn't designed — it emerges under real business pressure."
Smart TV SDK & Cloud IDE development.
"Cloud Native wasn't a buzzword here — it was built line by line."
Where it truly began — a solid engineering foundation was forged here.
"IDE, compilers, operating systems — the foundation of systems engineering was laid here."
Using probabilistic tools to verify probabilistic outputs equals no verification
When AI Agents are unpredictable, engineers shift from executors to tamers
Those who decompose vague requirements into precise specs are the scarcest AI-era talent
Deconstructing LLMs from first principles: Attention is a learnable weighted sum
The endgame of attention economy pivoting to ROI economy
你好，我是
Sr. Staff Engineer & Bar Raiser @ Coupang
当 Execution 变得廉价，Judgment 才是护城河——但没有 Ground Truth 的 Judgment，只是猜测
了解更多既能定义 What，也能交付 How。从零构建过基础设施、开源过 5000+ Star 的工具、用 AI 落地过生产级系统——横跨系统工程、移动端、AI 三个技术周期。
跨界思考是我的工作方式：用编译器的视角理解 LLM，用认知科学解读 AI 意识，用逆向投资的逻辑做技术决策。30+ 篇 AI 深度文章，核心主张——当 Execution 变得廉价，Judgment 才是真正的护城河。
"How 的上限，取决于 What 的精度。"
— Johnson
在 Coupang 落地 Autonomous AB Clean 等生产级 AI 系统，同时输出 30+ 篇深度思考。我关注的不是 AI 能写多少行代码，而是当 Execution 变得廉价后，Judgment 如何成为核心竞争力 。
在 Coupang 落地的 AI 工程实践——自主式 AB 实验清理系统，将 AI Agent 嵌入研发工作流，实现实验生命周期的自动化管理
从 Agent 架构设计到 Claude Code 实战，探索人机协作的最优范式。提出 Harness Engineering——在混沌中建立秩序的工程方法论
用概率性工具验证概率性输出等于没验证。AI 时代最被低估的竞争力是构建确定性验证体系
LLM 是函数而非数据库，从编译器到 LLM 的软件分层轮回——用第一性原理理解大模型
当 Agent 成为入口，注意力经济向 ROI 经济转型。App 退守后端基础设施，软件工程迎来最后一次范式革命
能把模糊需求拆解成精确规格的人，是 AI 时代最稀缺的人才。How 的上限取决于 What 的精度
从零构建全公司移动端基础设施，从零打造 R.LUX 奢侈品 App，用 AI 赋能工程效率。跨团队架构决策、性能瓶颈根因分析、技术债优先级判断；作为 Bar Raiser 独立于 hiring manager 把关每一个招聘决策。
"定义技术方向，守住人才标准——确保每一个进来的人都能拉高团队的平均水平。"
大规模移动端架构设计与性能优化。
"亿级 DAU 的短视频场景，极致体验优化没有捷径。"
主导司/乘移动端中台架构，开源 Booster 构建优化工具。
"好的基础设施不是被设计出来的，是在真实业务的压力下长出来的。"
Smart TV SDK 与 Cloud IDE 开发。
"Cloud Native 不是概念，是在这里一行一行写出来的。"
职业生涯的真正起点，扎实的工程基础在这里练就。
"IDE、编译器、操作系统——系统工程的底子，是在这里打下的。"