06 · About

Saurabh
Khanwani

Runs AICHROM from Delhi. An architect by training who followed form out of buildings and into real-time worlds.

Saurabh Khanwani, in a black turtleneck against a white arched backdrop
Delhi

Background

Saurabh trained as an architect and spent his first years on buildings, then on spaces that were never meant to be built: virtual environments, game-like interactive places, early metaverse concepts. That work raised a question he has been answering since: what does a space need to be when it only exists on a screen?

He took time out to learn computational design properly, then worked as an architect on parametric geometry for large commercial and interior projects in Rhino and Grasshopper. Geometry as logic, not drawing, is still how he thinks.

From there he moved into product design: 3D modelling, AI-generated graphics for landing pages and award films, web builds that started in Webflow and ended in custom code, motion in After Effects. Today he leads product work around computational design, building onboarding systems and interfaces with AI-assisted tools, and runs AICHROM alongside it.

Now

  • Worldsthree.js and GLSL, written with Claude Code as a pair
  • MotionGSAP for the page, Rive for the parts that answer back
  • NativeiOS experiments, taking worlds onto a device
  • FormRhino and Grasshopper when geometry needs to be exact

Reach

hello@aichrom.org
LinkedIn

Experiments

How the work gets made.

  1. E.01

    Claude as the pair

    Every line of this site, from the GLSL to the nginx config, was written with Claude Code in the loop: Saurabh directs, reviews and redirects; Claude writes and tests. The same workflow built a computational design awards site earlier this year, and it is how Claude-based automation found its way into his design process, from prompt-driven graphics to code-generated interaction.

  2. E.02

    GSAP for choreography

    Scroll reveals, pinned heroes, section colour transitions, word-by-word entrances that reverse on exit. GSAP and ScrollTrigger carry the motion on this site and on the awards site before it. The rule is that motion must say something about state, or it gets cut.

  3. E.03

    Rive for things that answer back

    State machines instead of clips: the interaction page shows the mechanism built by hand in canvas, and Rive is the tool for shipping it inside real interfaces, where a hover interrupted by a press has to blend instead of snap.

  4. E.04

    three.js, then the device

    Worlds start as shaders in the browser: raymarched terrain, SDF corridors, liquid chrome, point clouds, a walkable city. Then they go onto an iPhone to see what a battery, a thumb and sixty frames do to the design. The native experiments are where the next AICHROM work is heading.

Next

Studio