Education

B.Eng. in Architectural Engineering, Tongji University, 2019
M.Arch. in Architectural Design (Distinction), University College London, 2021
Ph.D. in Architecture (Design Computation), Georgia Institute of Technology, in progress (started Aug 2022)

Keywords

Computational Design, Parametric Design, Shape Grammars, Computer-Aided Design, Design Theory

Biography

Yichao Shi is a committed doctoral student in the Architecture Design Computation program at Georgia Institute of Technology. Yichao possesses a strong background in computational design, thanks to his Bachelor of Engineering degree in Architectural Engineering from Tongji University, his M.Arch in Architectural Design from University College London (UCL), and his Master of Science in design computation from Georgia Tech. His research interests include shape grammar in architectural design, digital heritage, computer graphics, cognitive science, and human-computer interaction. Yichao is an individual who embraces and utilizes AI tools for architectural design. He has conducted studies concerning the application of pix2pix GAN, Midjourney, Stable diffusion Web UI, and Comfy UI as tools for design assistance.

Statement of Teaching Interest

I am interested in teaching courses that connect architectural design with computation and AI. My teaching topics include computational design fundamentals, parametric modeling (Rhino/Grasshopper), shape grammars and rule-based generative systems, computer-aided design workflows, and AI-assisted design methods for early-stage exploration.

Statement of Research Interest

My research focuses on the intersection of shape grammar, AI and architectural design/technology, with an emphasis on rule-based and interpretable computational methods. I study shape grammars as a way to represent design knowledge, control generation, and connect formal design operations to computation. I am also interested in how modern AI models (such as generative models and learning-based predictors) can work together with explicit rules, so the system can produce designs that are both flexible and explainable.