Systems
I design systems, not just outputs. Each of these started with the same question:
what's the repeatable logic underneath this problem — and how do you build a pipeline around it.
[01]
BeyndAI
Turns raw dealership forecourt photos into studio-quality listing images — GPT-based generation, automated segmentation, and deterministic compositing replace a manual photo studio entirely.
[02]
Scientific Dataviz
An automated pipeline that converts molecular structure data into rendered visualizations — repeatable, accurate outputs without manual scene-building for every dataset.
[03]
Photogrammetry convertor
Converts photogrammetry scan data directly into Unreal Engine-ready assets — built for NYT, removing the manual cleanup step between capture and real-time render.
