Deepform

Links

The Problem

Architectural feasibility analysis required manually reading Rhino 3D models, a process that was slow, Windows-only, and impossible to run in a web browser or cloud environment.

What I Built

• Built DeepformCore, a .NET Core analysis engine that reads Rhino 3dm files via an AWS Lambda function on Linux — a significant departure from the Windows-only norm in AEC software.
• Classified arbitrary 3D geometry into architectural components (floors, programs, levels) and produced floor area breakdowns by program and floor for feasibility dashboards.
• Adopted Hypar Elements, an open-source cross-platform BIM library, for JSON-serializable geometry processing to support future file format expansion.
• Built the main application backend in NestJS/TypeScript with TypeORM, Redis, and BullMQ for background job processing.

Tech Stack