Regular mode
Focused, familiar browsing.
Navigate with your mouse, keyboard, or touch device. The portfolio remains complete and fully usable without camera access.
Interactive Vision Lab
Explore the site normally, or opt into a browser-local hand-tracking layer that lets you aim, scroll, and open work with natural gestures.
Interactive mode is optimized for desktop browsers with a camera and precise pointer input. Regular browsing is always available.
Two ways to explore
Regular mode
Navigate with your mouse, keyboard, or touch device. The portfolio remains complete and fully usable without camera access.
Interactive mode
Opt in when you want to see the computer vision layer in action. The interface is designed to complement—not replace—standard controls.
How it works
The interaction layer loads only when requested, keeping the default portfolio fast and conventional.
Nothing starts by default. Interactive mode asks for camera permission only after you choose to begin.
The browser reads the live video locally and estimates 21 hand landmarks, giving the interface a lightweight hand map.
Open hand aims, a closed fist scrolls, and a point or pinch activates the focused control. Gestures are deliberately limited and predictable.
Turn the camera off, hide the skeleton overlay, or exit at any time. Turning the camera off or leaving interactive mode stops the webcam stream.
Live controls
Once interactive mode is active, a compact panel appears in the upper-right corner. It makes the visual layer adjustable without interrupting navigation.
Technical note: this portfolio currently uses MediaPipe Tasks Vision’s hand-landmark model. It follows the same browser-side inference pattern commonly associated with TensorFlow.js: the model is loaded on demand, and video frames are processed in the browser rather than sent to this site.
LIVE DISPLAY