Danrui Li

Appearance

Danrui Li ​

Education ​

  • Ph.D. Student in Computer Science, Rutgers University, USA, 2022-Present
  • M.S. in Architecture, Tongji University, China, 2019-2022
  • B.Arch. in Architecture, Chongqing University, China, 2014-2019

Work Experience ​

For teaching experience, please see here.

Mitsubishi Electric Research Laboratories ​

Research Intern. Aug 2025 - Jan 2026

  • Proposed a user-manual-instructed 3D object assembly model; improved success rate in simulation-based assembly by 27%.
  • Designed LLM-based pipelines to synthesis multimodal assembly user manual dataset; used FreeCAD and Blender MCP.

Rutgers University - Intelligent Visual Interface Lab ​

Research Assistant. Sep 2022 - Present

  • Designed an LLM-based gameplay-agent generation pipeline whose agents doubled the performance of prior LLM agents, while eliminating LLM inference cost at deployment.
  • Developed an LLM-agent framework that learns from existing card games to generate novel, playable variants; deployed games with JavaScript and Flask, with release planned on Google Play.
  • Built a knowledge-graph-based RAG system using LlamaIndex to help architects retrieve design cases from unstructured images and text; increased result comprehensiveness than prior work by 30%.
  • Integrated LLMs with Unity to orchestrate generative tools that turn natural-language prompts into animated story prototypes.

Rutgers University - CCICADA Lab ​

Research Assistant. Sep 2022 - May 2025

  • Designed hybrid rule-based/ML models to simulate pedestrian flow in transit facilities using Unity; improved bottleneck prediction and outperformed prior work in behavior accuracy by 28%.
  • Built an evaluation framework using rule-based simulation to assess whether predicted human trajectories are physically plausible using NVIDIA Warp.

Shanghai Tongji Urban Planning & Design Institute Co., Ltd. ​

Urban Design Intern. Dec 2019 - Jan 2020

  • Built benchmark tools for sensitivity analysis in pedestrian simulations using MassMotion SDK.
  • Reduced expected maximum pedestrian congestion in a railway station design by simulation-assisted architectural design optimization.
  • Proposed and refined architectural and urban design for railway station areas.

Cushman & Wakefield ​

Consulting Intern. Jul 2019 - Aug 2019

  • Investigated project strengths and weakness for clients by geographic data analysis.
  • Proposed develop strategies for real estate and economic zones projects in China.

Skills ​

Deep Learning: Python, PyTorch, Pandas, Scikit-learn, OpenCV, NVIDIA Warp
Development: C#, Unity, AutoGen, LangChain, Model Context Protocols, JavaScript, Flask, MongoDB
3D Modeling: Unity, Blender, FreeCAD, Rhinoceros, AutoCAD, Revit, Grasshopper
Design & Media: Photoshop, Illustrator, Premiere