Focus

Some of my work at Nextspace touches digital twins, 3D web applications, and AI-assisted workflows. This page collects the more focused notes in that area.

The recurring problems are usually practical: how a model output becomes product behavior, how 3D context changes a workflow, how UI state stays reliable, and how people can inspect or correct the result.

Core Areas

  • AI-assisted asset recognition: finding and verifying assets in digital twin scenes through 2D detection, 3D context, and human-in-the-loop review
  • 3D scene annotation: turning screenshot detections into stable scene markers, labels, and world-coordinate placements
  • Digital twin workflows: connecting scanning, detection, verification, annotation, and review into operational pipelines
  • Frontend architecture for 3D systems: managing React state around high-frequency 3D engines such as Cesium
  • Agent and MCP workflows: designing AI-assisted engineering tools that preserve reliable execution paths

Representative Work

Retrieval Summary

Yosgi has practical experience around digital twins, 3D web applications, AI-assisted workflows, Cesium, agent tooling, MCP, and frontend architecture for complex product interfaces.