Lead Visual Designer
San Francisco
Currently @Creatify AI

LinkedIn Instagram

Wahaj: light that notices you

Wahaj (وهج, Arabic for glow) is an interactive light and sound installation that nudges people into noticing each other. The more people gather under it, the brighter it gets.
Visitors under the glowing tubes at the exhibition
The Wahaj poster
Abed at Wahaj

Design

  • Concept and research
  • Spatial and light design
  • Poster and identity

Build

  • Person detection with a camera
  • Soldering, Arduino and LEDs
  • 3D-printed end caps
  • Sound and chimes

Advisors

  • Prof. Alex Baumgardt
  • Prof. Graham Plumb

The question

How do you get strangers to notice each other?

How might we create a physical, interactive space that encourages college students to break social barriers and engage with one another through shared sensory experiences?

Campus is full of people walking past each other. I wanted a space that rewards slowing down and standing together, without asking anyone to tap a screen or download anything.

The pitch

Illuminating connections.

Abed presenting the Illuminating Connections opportunity analysis
The pitch, Oct 2024

I pitched the idea in October 2024 as an opportunity analysis called Illuminating Connections. The research started with installations that turn people into the input: Rafael Lozano-Hemmer's Pulse, which turns heartbeats into shared light; Squidsoup's Submergence, where movement shapes a room of points; and Charles Gadeken's Entwined, a glowing tree visitors can recolor.

All three work because the crowd is the content. None of them needs instructions.

Pulse, Submergence and Entwined

The concept

Light that thrives on a crowd.

Suspended tubes of light hang over a common space. Empty, they rest in a soft warm glow. As people walk in, each tube responds, and the light intensifies with every extra person, so the installation literally thrives on community.

Concept render: tubes resting, nobody around
Concept render: tubes glowing as people walk through

First prototype

One tube, called Lumos.

By the end of fall 2024 I had one working tube. It was called Lumos then, and it hung by the Hybrid Lab with a poster laying out the five behaviors I was designing for: illuminate, react, intensify, adapt and inspire.

One tube was enough to test the things that matter: how bright it should be, how fast color should move, and how a room reacts when a light starts paying attention.

Abed next to Lumos, the first tube, at the end of fall 2024

Lumos at its first showing, and the prototype cycling through color in the studio.

Building it

Nine tubes, by hand.

Every tube is a strip of 82 addressable LEDs, cut to length, soldered and run back to an Arduino. I soldered all of it myself, then wired and tested each strip on a breadboard before it went inside a tube.

The end caps are 3D-printed in clear filament. The first ones had hard, flat ends, and feedback on them pushed me toward something gentler, so the final caps are rounded and soften the bottom of each tube. Then the tubes went up, one at a time, on a metal grid above the floor.

Soldering wires onto the LED strips
Measuring an LED strip against the length of a tube
An LED strip wired to an Arduino on a breadboard
3D-printed end caps: the rounded final version next to an earlier flat one
The first finished tube lit up in the empty exhibition hall
Hanging the tubes from the metal grid

Soldering, sizing, wiring, end caps, the first finished tube, and hanging the rest.

Teaching it to see people

A camera that looks straight down.

The hardest part was sensing people reliably. I tried different sensors, then landed on a camera running a person-detection model. I built the pipeline in Python, ran detection through Roboflow, and connected the results to the lights.

Then I hit a wall. The camera hangs above the installation looking straight down, and the general model was trained on people seen from the front. From above, it kept missing them. So I retrained it: I took photos from the camera's own angle, pointed down at people, labeled them, and trained the model on that view.

93% detection success after retraining on the top-down view.

An early test, before the camera went overhead.

The overhead camera's view of the floor on a laptop
Training batch of top-down camera photos with predicted boxes

Setting up the camera on site, the top-down training set, and checking the live feed.

How it works

Nine lanes, nine tubes.

Nine tubes of 82 LEDs hang from a metal grid, with a speaker on each side. A camera above splits the floor into nine lanes, one under each tube, and tells the lights who is standing where.

Empty, the room breathes in soft pastels. Once someone walks in, the light follows people: your tube blooms, the empty ones fade, and two people apart get a bridge of light between them. Turn the sound on to hear it too.

Breathe: With nobody around, all nine tubes breathe in soft, drifting pastels over a quiet ambient loop.

At the exhibition

Strangers, working it out together.

Wahaj ran at the CCA Interaction Design graduating exhibition in May 2025. People walked up curious, then stayed to figure out what made it change, and the best moments were strangers working it out together.

Abed at Wahaj on the exhibition floor
Visitors under the glowing tubes
Abed walking a visitor through Wahaj, the tubes glowing in a rainbow behind them
Guests gathered beside Wahaj on the exhibition floor
Wahaj thesis poster