research

The Problem

Generative design has revolutionized how engineers explore design possibilities, using AI and optimization to create solutions that push beyond traditional constraints. Yet a fundamental gap remains:

Most generated designs cannot be manufactured at scale. The intricate geometries that emerge from topology optimization and deep generative models often require additive manufacturing—a process poorly suited for mass production.

This leaves industry with beautiful but impractical designs that require extensive manual rework before they can reach the factory floor.

My Approach

My research bridges this gap by embedding manufacturing constraints directly into the generative design process. Rather than treating manufacturability as an afterthought, I develop methods that produce production-ready designs from the start.

A key insight driving my work is that complex 3D manufacturing constraints can be elegantly captured through 2D depth representations. This simplification makes AI-based design generation both computationally tractable and directly aligned with processes like die casting and injection molding, where part geometry must respect mold separation and draft angles.

By working in 2D, we can leverage powerful image-based generative models while ensuring every output is inherently manufacturable.

Philosophy

Engineering research must serve industry to be meaningful. A paper that cannot translate to the factory floor—no matter how novel—has limited value.

This conviction shapes my work: every method I develop is tested against real manufacturing requirements, and many have been deployed in industry projects through Narnia Labs.

Current Focus

My research addresses several interconnected challenges:

Structural Components

• Brackets, supports, and load-bearing parts optimized for die casting and injection molding
• Ensuring generated geometries respect draft angles, wall thickness, and mold separation

Mechanism Design

• Four-bar linkages and kinematic systems synthesized using deep generative models
• Balancing kinematic requirements with quasi-static performance

Manufacturing Processes

• Die casting and injection molding constraints embedded in generative frameworks
• Exploring extensions to other mass-production methods

Future Directions

Looking ahead, I aim to expand these principles beyond two-mold processes to a broader range of mass-production methods. The goal is not just to constrain generative design, but to make manufacturing-aware AI the default—enabling engineers to explore vast design spaces while knowing every candidate can be built.