How I Turn Learning Into a Workflow in the AI Era 【2026-04-06】Before AI, learning was slow but joyful. Now I treat it like a workflow: collect, filter, outline, draft, refine, publish. Here is how that process actually works.
You Don't Know LLM Training: Principles, Pipelines, and New Practices 【2026-04-03】Most people think model upgrades mean bigger parameters. But the real differences you feel day-to-day usually come from the back half of the training pipeline. This piece follows the full chain from pretraining through distillation and deployment, with a focus on how data engineering, system recipes, post-training, reward design, and agent training together shape the final product. The conclusion is that models getting stronger is almost always a joint effect of weights, training pipeline, and deployment decisions, not just scale.
The Death of the Manual Programmer 【2026-03-30】The title is borrowed from a rewritten version of a song by Omnipotent Youth Society that I loved more than a decade ago. The scene in that song is obviously different, but the feeling of watching a familiar world slowly get replaced feels strangely similar.
You Don't Know AI Agents: Principles, Architecture, and Engineering Practices 【2026-03-21】After writing my deep dive on Claude Code's architecture, I realized my understanding of the underlying agent foundations was still lacking. Given our team's growing experience deploying agents in production, we desperately needed a systematic overview. I revisited the literature, open-source implementations, and my own code to comprehensively cover control flow, context engineering, tool design, memory, evaluation, and security.
You Don't Know Claude Code: Architecture, Governance, and Engineering Practices 【2026-03-12】This handbook distills six months of hands-on Claude Code usage across two accounts. It covers context management, Skills, Hooks, Subagents, prompt caching, and CLAUDE.md design, with a focus on making agent collaboration stable, governed, and verifiable.
Installing OpenClaw Isn't the Same as Using It 【2026-03-07】Watching the frenzy around Tencent Tower installing OpenClaw today gave me a lot to think about. Many big tech companies are aggressively pushing non-technical frontline employees to install this AI tool, with some even offering 500 RMB door-to-door installation services. Everyone's desperately searching for use cases, demanding implementation, and trying to prove this thing is too important to miss. The whole process gives me a strong sense of cyber-tech folding.
Bitcoin's Fall: Cathedral vs Casino 【2026-02-01】When Bitcoin falls, it becomes easier to distinguish between the casino and the cathedral. As Bitcoin dropped from its $120k peak to around $70k, panic swept through the market once again. During previous major stock market crashes. Price volatility is the casino's noise, while what truly determines long-term returns are those cathedrals that require a century of devotion yet continuously create structural value.
Best Purchases from 243 Engineers in the Past Year 【2026-01-24】Last week, while shopping for New Year's Eve gifts for my team, I casually asked on X what the most satisfying purchase people made in the past year was. I received 243 responses, discovering countless great products ranging from essential tech to practical everyday items.
Next-Gen Engineers: Breakthrough and Growth 【2025-12-22】Recently, I shared "Breakthrough and Development of the Next Generation of Engineers - Practical Transformation from Position to Capability" at AICon in Beijing. I have converted the PPT into images and posted them here, looking forward to exchanging ideas with you.
The Impact of AI Coding on Programmers 【2025-08-17】After using Claude Code for less than a month and spending $326 (actually $20 for Pro + $50 for top-up), the Cursor I had been using for a few months has become obsolete. Using AI well can easily help you reach the level of a senior engineer. As an engineer, I feel both amazed and concerned.