I'm an explorer bridging the digital and physical worlds. I enjoy building things that make sense of complex data, whether it's raw pixels, dense research papers, or deep debugging sessions.
An AI-agent system that bridges LLMs with Obsidian to automate deep research and personal knowledge management. It helps me maintain a high-velocity learning loop by automating the capture and synthesis of information.
A VS Code extension designed for computer vision engineers. It visualizes 1D/2D/3D data structures directly during C++ debugging sessions, turning abstract memory buffers into intuitive images and plots.
A handy tool to automatically generate and inject clickable tables of contents into PDF documents. Perfect for long technical reports or reconstructed eBooks.
A deep reinforcement learning project where a Deep Recurrent Q-Network (DRQN) learns to navigate and solve environments in MazeWorld.
My focus areas include:
- Computer Vision: Making machines see and understand.
- Edge Intelligence: Pushing heavy models to run fast on light hardware (TensorRT/ONNX).
- AI Tooling: Using AI to build better tools for humans.
- 🌍 Based in Hong Kong
"Pushing the physical limits of hardware to operationalize human dignity."