A final-semester UQ student. I use AI to build small tools and run research for myself — a few are below, including one you can open and try.
A single-file web app I designed and built for UQ students — a curated guide to the quietest study spots (which floors are calm, and when), with live room-occupancy as the next layer I'm wiring in. A working prototype.
Open the live app →It's running right here. Scroll inside it, or open it full-screen above.
Each runs on its own. Some I built from scratch, some I wired together from open-source — all of them I operate daily. (No engineering background.)

A skill I built for my AI assistant — one reusable command. Point it at a topic and it drives a real phone to search Xiaohongshu (China's biggest discovery app), reads the posts, and writes the findings into my notes.
Fast scan or deep dive. The screenshot is it mid-run — searching, then auto-summarising 55 posts.
“Both sides restart high-level talks; a leaders' meeting yields mutual tariff cuts — but tariffs stay the core bargaining chip.”
An open-source prediction engine that turns the day's news into a structured forecast. I run it on a daily automation — it briefs itself every morning while I sleep, and posts the result.
Engine by 666ghj/MiroFish. The daily pipeline around it is mine.
My personal operations desk: five AI agents that research, draft, trade and report on their own — around a dozen scheduled jobs, with failure alerts and self-recovery.
Built on the open-source OpenClaw runtime. The agents, routing and schedules are mine.