I have a soft spot for turning wild ideas into things that actually exist.

I finished AP Micro with a 5. What actually blocked me wasn't the economics — it was reading English questions and figuring out what they were really asking. So EasyEcon keeps the English originals and gives Chinese students scaffolding in Chinese context to understand them. I wrote the question bank, designed the flows and the mock-exam experience, and vibe-coded it into being. The mock mirrors the College Board Bluebook — timed MCQ and FRQ, with 2025 real papers. A one-page brief to our international-office director won a ~40-student Grade-10 pilot this term.

On a four-person team I owned product direction and user scenarios, plus the pitch narrative and visual assets. Instead of yet another vocab app or AI chatbot, we put the scenarios inside high-pressure real conversations — interviews, service counters — where you face a "person" who actually pushes back. Online you use the app; in-person it can pair with AR glasses for live translation and answer cues — the hardware is never required. ≈8,975 applicants, 800 on site; we won the Ant Group Lingguang track's "Most Sincere Award" (¥2,000 team prize). It was the first time I clearly realized: no matter how beautifully a product is pitched, without a real person and a real scenario, it's still just a PowerPoint.

In September 2025, just starting to learn vibe coding, I built the IELTS Speaking tool I wished existed — full Part 1 / 2 / 3 mock exams and scoring, in a flow close to the real test. About 32 people used it. But "can be built" and "worth continuing" turned out to be two different things: acquisition was too expensive and competitors were already strong, so I paused. It was the first time I truly felt, from inside my own product: a product can be pretty good and still not deserve to continue.