Hi, my name is
Mandy Dai.
I turn ideas into scalable code.
I’m a CMU graduate who enjoys building systems that combine strong technical foundations with real-world impact — from backend infrastructure to interactive AI-powered platforms.
About Me
Hey there! I’m Yuchen (Mandy) Dai — I completed both my B.S. and M.S. in Electrical and Computer Engineering at Carnegie Mellon (plus an additional major in Psychology). These days, I’m a research assistant atCarnegie Mellon AI-SDM Lab, where I help turn AI research into real, usable systems — from software engineering and data pipelines to ML integrations.
I like working across the stack and making complex ideas feel structured. Whether I’m designing backend architecture or refining a user-facing workflow, I care about building systems that are clean, reliable, and built to last.
So far, I’ve had the chance to work in a research lab, an AI-driven startup, a data analytics firm, which has given me exposure to different problem spaces — from simulation and interactive systems to production web applications. What ties it all together is that I like building systems that are thoughtful, reliable, and actually useful.
Here are a few technologies I’ve been working with recently:
- Python
- C#
- React
- Typescript
- Node.js
- Django
- Unity
- PyTorch
Where I’ve Worked
Research Assistant @ CMU AI-SDM Lab
Oct 2024 - Present
- Led collaboration with American Red Cross (ARC) stakeholders and CMU faculty to conceptualize and develop an AI-powered disaster simulation software for training purpose, integrating LLM-driven agents for real-time instructional guidance and decision-making.
- Integrated a client–server data collection pipeline in collaboration with the ML team to log user decisions and simulation states, enabling quantitative evaluation and AI-driven performance analysis.
- Built an admin-facing configuration system to decouple scenario logic from gameplay mechanics, enabling real-time parameter adjustments and eliminating code-level changes for experimental setup.
- Translated Figma designs into interactive Unity UI components, improving usability and consistency across training scenarios.
Some Things I’ve Built
Featured Project
ARC Disaster Simulation
At CMU’s AI-SDM Lab, I’m leading the development of an AI-powered disaster simulation platform for the American Red Cross — designed as both a controlled research environment and scalable training system, with fully configurable maps and scenarios, LLM-driven agents, and an integrated data collection pipeline for rigorous experimental evaluation.
- Unity
- C#
- JSON
![ARC Disaster Simulation]()
Featured Project
Cyber Imposter: AI Among Us
I contributed to Cyber Imposter: AI Among Us by developing the in-game log system, facilitating communication between the server and users, and also assisted with launching their official website.
- Unity
- C#
- Figma
- Framer
![Cyber Imposter: AI Among Us]()
Other Noteworthy Projects
view the archiveCLIP-UIQA
We built CLIP-UIQA, A multi-branch neural network for unified image quality assessment using CLIP Trained on a custom dataset with multi-dimensional subjective annotations (e.g., happy–sad, clear–blurry), CLIP-UIQA outperforms baselines in both convergence speed and perceptual alignment.
Microgrid Simulation and Optimization Tool
I built a web-based microgrid simulation tool that integrates machine learning-based forecasting and dynamic power flow visualization to help users design, optimize, and analyze renewable energy systems.
Finding Alex - Online Game
Big fan of Undertale-style games, so my friends and I speedran this creepy-cute mystery in 3 days. You play as Zoe, returning home from school only to find her best friend missing. Puzzles, suspicious adults, and just the right amount of unsettling energy.
What’s Next?
Get In Touch
I’m exploring new roles where I can continue growing as an engineer and building things that matter. Feel free to drop a message — I’d love to connect!
Say Hello