Served as a TA at the DIGITALFUTURES 2026 Workshop at Tongji University, and presented a research paper at the CDRF 2026 Conference (Sydney, Australia).担任同济大学 DIGITALFUTURES 2026 Workshop 助教，并在 CDRF 2026 Conference（澳大利亚悉尼）报告研究论文。
News动态
Published a research paper at the CAADRIA 2026 Conference (Hsinchu, Taiwan).在 CAADRIA 2026 Conference（台湾新竹）发表研究论文。
Participated in the Future Lab exhibition at the Shanghai West Bund Art Center as a core creator.作为核心创作者参与上海西岸艺术中心 Future Lab 展览。
Participated in the SFC competition held in Zhangjiakou and connected with design and construction units on-site.参与张家口 SFC 竞赛，并与现场设计及施工单位进行对接。
Personal Information个人信息
Zizheng Yu于子正
Master's Student in Architecture建筑学硕士研究生
Soochow University苏州大学
Birth Year: 2001出生年份：2001
Birthplace: Shenyang, Liaoning出生地：辽宁沈阳
Education教育背景
M.Arch. in Architecture (Recommended Admission)建筑学硕士（推免）
Research Focus AI-assisted construction design | Robotic fabrication | 3D printing | Human-machine collaboration研究方向 AI 辅助建造设计 | 机器人建造 | 3D 打印 | 人机协作
Soochow University - Suzhou, China苏州大学 - 苏州，中国
B.Arch. in Architecture建筑学学士
Shenyang Jianzhu University - Shenyang, China沈阳建筑大学 - 沈阳，中国
Project Experience项目经历
AI Product ManagerAI 产品经理 | ArchPrompt
Independent Project · 2026独立项目 · 2026
- Identified a key user pain point in AIGC workflows: users often struggle to translate vague visual intentions into precise, controllable prompts; defined ArchPrompt as a low-input-cost prompt optimization product.识别 AIGC 工作流中的核心用户痛点：用户常常难以将模糊的视觉意图转译为精确、可控的提示词；由此将 ArchPrompt 定义为低输入成本的提示词优化产品。
- Designed the end-to-end product flow from user intent, task classification, rule injection, optimized prompt, image generation, to evaluation feedback.设计从用户意图、任务分类、规则注入、优化提示词、图像生成到评估反馈的端到端产品流程。
- Built a 100-case benchmark covering different task types, input clarity levels, visual styles, and user constraints, and established a structured evaluation framework for model output quality.构建包含 100 个案例的评测基准，覆盖不同任务类型、输入清晰度、视觉风格与用户约束，并建立结构化的模型输出质量评估框架。
- Led two iterations of the Prompt Skill based on bad-case analysis, introducing intent fidelity, task-type routing, viewpoint preservation, and constraint control to reduce model hallucination and output drift.基于 bad-case 分析主导 Prompt Skill 两轮迭代，引入意图保真、任务类型路由、视角保持和约束控制机制，以降低模型幻觉和输出偏移。
- Validated the solution across representative scenarios, increasing output usability from 41.7% without specialized rules to 100% with the refined Skill v2.在代表性场景中验证方案效果，使输出可用性从无专项规则时的 41.7% 提升至 refined Skill v2 下的 100%。
AI Product DevelopmentAI 产品开发 | Structural FEA
Independent Project · 2026独立项目 · 2026
- Translated a complex engineering workflow into a functional AI-assisted product prototype, defining the end-to-end user journey from CAD import and model configuration to structural analysis, visualization, and export.将复杂工程工作流转译为可运行的 AI 辅助产品原型，定义从 CAD 导入、模型配置到结构分析、可视化与导出的端到端用户路径。
- Decomposed domain requirements into modular capabilities including CAD parsing, structural modelling, FEA solving, result interpretation, and file interoperability, and iteratively drove implementation through AI-assisted development.将领域需求拆解为 CAD 解析、结构建模、FEA 求解、结果解读与文件互操作等模块化能力，并通过 AI 辅助开发持续推进实现。
- Designed a CAD-to-FEA workflow supporting 3DM, DXF, and DWG, automatically converting line-based geometry into structured analytical models and reducing repetitive manual model reconstruction.设计支持 3DM、DXF 与 DWG 的 CAD-to-FEA 工作流，自动将线性几何转化为结构化分析模型，减少重复性的手动模型重建。
- Built support for 3D truss and frame analysis, multiple load cases, self-weight, 6-DOF constraints, and engineering outputs including displacement, reactions, stress, strain, axial force, shear, and bending moment.支持三维桁架与刚架分析、多荷载工况、自重、六自由度约束，以及位移、支座反力、应力、应变、轴力、剪力和弯矩等工程输出。
- Packaged the workflow into an offline Windows product prototype with modular frontend, local API, solver, and CAD-processing architecture.将工作流封装为离线 Windows 产品原型，形成模块化前端、本地 API、求解器与 CAD 处理架构。
Teaching Assistant助教 | DigitalFUTURES 2026 International Workshop — AI-assisted Design & Material Intelligence | DigitalFUTURES 2026 国际工作坊 — AI 辅助设计与材料智能
2026.06 - 2026.07
- Supported an interdisciplinary workshop integrating multimodal AI generation, parametric modelling, and digital fabrication into an end-to-end design workflow.支持跨学科工作坊，将多模态 AI 生成、参数化建模与数字建造整合为端到端设计工作流。
- Guided students in applying AI for concept generation, design iteration, and solution comparison, translating generative outputs into controllable digital models.指导学生运用 AI 进行概念生成、设计迭代与方案比较，并将生成结果转译为可控的数字模型。
- Mentored Team 3 — The Living Envelope from AI-assisted exploration to physical prototyping, developing a modular system combining 3D-printed structures with natural materials.指导第三组 — The Living Envelope 从 AI 辅助探索推进到实体原型制作，发展结合 3D 打印结构与自然材料的模块化系统。
Honors & Awards荣誉与奖项
2026
- Teaching Assistant Recognition, DigitalFUTURES 2026 — AI & Material Intelligence WorkshopFrom Vision to Matter: Reprogramming Architecture in the Age of AI and Material Intelligence助教表彰，DigitalFUTURES 2026 — AI 与材料智能工作坊从愿景到物质：AI 与材料智能时代的建筑再编程
2025
- Second Prize for Excellent Graduate Student, Soochow University Graduate Academic Scholarship苏州大学研究生学业奖学金优秀学生二等奖
2024
- Academic Excellence Scholarship, Shenyang Jianzhu University沈阳建筑大学学习优秀奖学金
2023
- Second-Class Scholarship, Shenyang Jianzhu University沈阳建筑大学二等奖学金
2022
- Student Member, Architectural Society of China中国建筑学会学生会员
Academic Research学术研究
Conference Paper会议论文
"ForkTension: Human-in-the-Loop Fabrication of a Hemp-timber Tensegrity Structure via Gesture-based Material Analysis and Robotic Weaving"
"Human-In-The-Loop As-Built Structure Analysis Through Motion Capture and Virtual Reality"
Affiliated with隶属于
Humachine Lab
SAUP- Skills:技能： Vibe Coding, Robotic Fabrication, 3D Printing, PhotographyVibe Coding，机器人建造，3D 打印，摄影
- Tools:工具： Codex, Claude Code, Photoshop, Illustrator, Grasshopper, Rhino, Lumion, AutoCAD, SketchUpCodex，Claude Code，Photoshop，Illustrator，Grasshopper，Rhino，Lumion，AutoCAD，SketchUp
- Hobbies:爱好： Films, Basketball, Games, Music电影，篮球，游戏，音乐