煜煜华inxpla-engineering.hashnode.dev·Sep 7 · 7 min readHow I Make AI API Docs Useful Before JavaScript LoadsI used to treat the documentation shell and the documentation content as the same thing. If the page looked complete after hydration, I considered it done. That was too optimistic. An API documentatio00
煜煜华inxpla-engineering.hashnode.dev·Sep 4 · 5 min readA Shot Queue for Veo 3.1 Product VideosI would not give one AI video task responsibility for product identity, a mechanism demonstration, a lifestyle scene and the finished edit. The problem is not only output quality. A large prompt produ00
煜煜华inxpla-engineering.hashnode.dev·Sep 2 · 4 min readA Request State Machine for GPT Image 2 Product ImagesThe most expensive image API bug is not always a failed request. Sometimes it is a request that succeeds twice. I saw why a normal “retry” button is too vague for a GPT Image 2 API workflow: a timeout00
煜煜华inxpla-engineering.hashnode.dev·Sep 2 · 3 min readHow I Design Comparable TikTok Shop Product SnapshotsWhen I build a product-research workflow, the most dangerous bug is not always a failed request. It is a successful request whose output gets mislabeled. A cumulative sales field can be useful. It bec00
煜煜华inxpla-engineering.hashnode.dev·Sep 1 · 9 min readUnified AI API Tutorial: Keep Model Contracts ExplicitWhy I wrote this When I review a unified AI API, I do not start by counting model names. I start by checking which contracts are explicit, which jobs are asynchronous, and where the application must k00
