AR-02 · PROJECT ARCHIVEEngineering Workshop
Things that were built, with the reasoning kept in. Each record separates what the repository reports from what its author says about it — the first is fetched, the second is written. Where a project’s interesting part is how its intelligence works rather than what it does, that is taken apart in the AI Lab.
| No. | Project | Class | State | Language | Last worked |
|---|---|---|---|---|---|
| 001 | YushaCyber | Platform | active | Python | |
| 002 | Jarvis Assistant | Desktop assistant | active | Python | |
| 003 | AgroVision Nepal | Simulator | paused | HTML |
001Platformactive
A cybersecurity learning platform built as a Flask application — roadmaps, labs, a CTF arena, an AI mentor and a browser terminal, wired together as fourteen registered blueprints.
002Desktop assistantactive
An offline-first desktop assistant that hears, sees and acts — local wake word, local transcription, local LLM, and hand-gesture control, with nothing sent to a cloud API.
003Simulatorpaused
A Flask agriculture simulator that identifies crops from a photograph by mapping a general-purpose ImageNet classifier onto five crops grown in Nepal, then scores whether your soil and this week's weather suit the result.