// PROJ_01AI / LLMALL PROJECTS_
The complete index — 23 projects on 5 shelves.
Pick a shelf or search the stack.
Applied Work
09Research, client and coursework builds — AI pipelines, vision systems, mobile apps.
// PROJ_01AI / LLM
// PROJ_02Computer VisionParcel/Package Delivery Detection System
Parcel detection is a specific application of object detection, where the goal is to identify and locate parcels (packages) within images and match it with sets of condition for proof of delivery (POD)
// PROJ_03AR / Mobile
// PROJ_04Computer VisionPortable Intruder Detector System
Achieved 98.7% sub-2.3s detection using ML and image processing to detect human and animal intruders.
// PROJ_05AR / MobileAR Travel Application - Memorable
Travel app exploring landmarks through NFC-triggered content, ARCore interactions, and AI-powered chats.
// PROJ_06Web / FrontendGDXChange — AI-Supported Security & Infrastructure
AI-supported security and infrastructure website for GDXChange, built as a responsive, fast-loading frontend.
// PROJ_08AR / Mobile
// PROJ_18Web / FrontendChorus — Live Q&A For A Whole Room
A presenter authors a deck of questions and runs it in front of a room while the audience answers from their phones — five question types rendering live on a stage view, with server-authoritative timing, scoring and a top-five leaderboard.
// PROJ_23AI / LLMlark-acp-bridge — Run Any AI Coding Agent From a Lark Chat
An open-source bridge that turns a Feishu/Lark bot into a front end for any Agent Client Protocol agent: Claude Code, Kiro, Codex, Gemini CLI, Amazon Q and more. Messages arrive over Lark's WebSocket, so it runs on a laptop behind NAT with no public endpoint, and the agent's thinking, tool calls and answer stream into one interactive card, with per-tool permission prompts and sessions that survive a restart. Extended from 4t145/lark-acp with Kiro and Amazon Q support, Lark/Feishu region switching, Windows builds, private-by-default access control (owner, admin, user and group allowlists), and a Lark MCP tool server that lets the agent ask you a question with a card or pick up a file you attached. Published on npm.