I'm Abdullah, a product engineer. Most of my work lives in the gap between what a product should be and what it takes to ship it: direction, architecture, and the follow-through to get both into production.
Lately that means AI products, developer tools, and the web platforms behind them. I care less about clever tech than about software that holds up: users understand it, and the team can keep building on it.
Some products need their first working version. Others need someone to untangle what's already there. Either way, I start from the real constraints and ship in small steps you can see.
void main() {
vec4 pos = texture2D(uPositions, vUv);
vec4 info = texture2D(uInfo, vUv);
vec2 mouse = uMouse;
float radius = length(pos.xy) * 1.0;
float circularForce = 1. - smoothstep(
0.3,
1.4,
pos.x - radius
);
float angle = atan(pos.y, pos.x)
- info.y * 0.3 * mix(0.5, 1., circularForce);
float targetRadius = mix(
info.x,
1.0,
0.5 + 0.45 * sin(angle * 2. + uTime * 0.6)
);
radius += (targetRadius - radius)
* mix(0.2, 0.5, circularForce);
vec3 targetPos = vec3(
cos(angle),
sin(angle),
0.0
) * radius;
pos.xy += (targetPos.xy - pos.xy)
* 0.01
* uMotionSpeed
* uFrameScale;
pos.xy += curl(
pos.xyz * 4.,
uTime * 0.1,
0.1
).xy * 0.0005 * uMotionSpeed * uFrameScale;
float distToMouse = length(pos.xy - mouse);
vec2 dir = normalize(pos.xy - mouse);
if (uMouseClicked) {
float influence = smoothstep(
0.5,
0.0,
distToMouse
);
vec2 tangent = vec2(-dir.y, dir.x);
float spin = info.y * 2.0;
pos.xy += tangent * 0.005 * spin * influence * uFrameScale;
pos.xy -= dir * 0.001 * influence * uFrameScale;
} else {
pos.xy += dir * 0.01 * uFrameScale * smoothstep(
0.5,
0.0,
distToMouse
);
}
gl_FragColor = vec4(pos.xy, 1.0, 1.0);
}Titles only tell you so much. What matters is what got faster, clearer, or possible at each place.
I ship customer-facing features across TanStack Start, Next.js, Django, and Azure, and help set technical direction along the way.
Led product engineering for an AI workflow product. Most of the job was making a complex tool feel fast and behave predictably.
Took an education product from nothing to production: the app itself, the AI orchestration behind it, and the cloud infrastructure under it.
Worked on analytics and data processing for high-volume financial workflows.
A few projects I keep coming back to. Each one shipped, and each one moved a number that mattered.
A commissioned social app that maps your connections as an interactive bubble field. I built it end to end, from the first sketch to the delivered product.
Research into making 3D reconstruction from photos faster and cleaner: a custom BRISK–SIFT feature extractor, CUDA acceleration, and better outlier removal.
The stack changes with the problem. These are the areas I keep coming back to.
If you're shaping a hard product, or you have a system that's outgrown its design, I'd like to hear about it. Send me the constraints.
EMAIL
LINKEDIN
GITHUB
X / @KHADELIDOTAI
Based between Doha and Edmonton.