Yu-Tung Liu

Research:
I am a first-year PhD student at the University of Maryland, College Park, under the supervision of Prof. Cunxi Yu. My research focuses on building autonomous AI agents for chip design automation, leveraging LLMs to drive the full ASIC flow (RTL generation, synthesis, physical design) with human-in-the-loop oversight. Previously, I developed neural methods for IR drop estimation at NYCU (1st place, ICCAD 2023 CAD Contest) and applied diffusion and state-space models to biomedical signal denoising at Academia Sinica.

Education: I obtained my Bachelor’s degree in Electronics and Electrical Engineering from National Chiao Tung University (NCTU), where I worked with Prof. Hung-Ming Chen and Prof. Juinn-Dar Huang. I was a research intern at Academia Sinica, where I worked on efficient learning algorithms for biomedical signal processing with Dr. Yu Tsao.

news

Sep 24, 2026	Our paper Structured Human-Like Agentic Flow for RTL Design has been accepted to NeurIPS 2026!
Jul 29, 2026	VeriTrace will be presented at ICLAD 2026 as a long oral!
Jul 25, 2026	I am honored to join the Young Fellow program at DAC 2026.
Mar 11, 2026	Launched Abstracted, a free PWA for researchers! Feel free to play with it.
Mar 11, 2026	I will be presnting a poster at NSF Workshop on Agents for Chip Design Automation!

selected publications

(*) denotes equal contribution

NeurIPS

Structured Human-Like Agentic Flow for RTL Design

Yu-Tung Liu, Zhan Song, Chenhui Deng, and 2 more authors

In The Fortieth Annual Conference on Neural Information Processing Systems (NeurIPS), 2026
ICLAD

VeriTrace: Human-Like Temporal Exploration Completes Agentic Action Space

Yu-Tung Liu, and Cunxi Yu

In IEEE International Conference on LLM-Aided Design (ICLAD), 2026 (Long Oral: Top 20%)

PDF
DAC

TOPCELL: Topology Optimization of Standard Cell via LLMs

Zhan Song, Yu-Tung Liu, Chen Chen, and 6 more authors

In 2026 63rd ACM/IEEE Design Automation Conference (DAC)

PDF
ICCAD

CFIRSTNET: Comprehensive Features for Static IR Drop Estimation with Neural Network

Yu-Tung Liu^*, Yu-Hao Cheng^*, Shao-Yu Wu, and 1 more author

In IEEE/ACM International Conference on Computer-Aided Design (ICCAD), 2024

Abs PDF Slides

IR drop estimation is now considered a first-order metric due to the concern about reliability and performance in modern electronic products. Since traditional solution involves lengthy iteration and simulation flow, how to achieve fast yet accurate estimation has become an essential demand. In this work, with the help of modern AI acceleration techniques, we propose a comprehensive solution to combine both the advantages of image-based and netlist-based features in neural network framework and obtain high-quality IR drop prediction very effectively in modern designs. A customized convolutional neural network (CNN) is developed to extract PDN features and make static IR drop estimations. Trained and evaluated with the open-source dataset, experiment results show that we have obtained the best quality in the benchmark on the problem of IR drop estimation in ICCAD CAD Contest 2023, proving the effectiveness of this important design topic.
TCAD

GIRD: A Green IR-Drop Estimation Method

Chee-An Yu, Yu-Tung Liu, Yu-Hao Cheng, and 3 more authors

IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 2025

Abs PDF

An energy-efficient high-performance static IR-drop estimation method based on green learning called GIRD (Green IR Drop) is proposed in this work. GIRD processes the IC design input in three steps. First, the input netlist data are converted to multi-channel maps. Their joint spatial-spectral representations are determined with PixelHop. Next, discriminant features are selected using the relevant feature test (RFT). Finally, the selected features are fed to the XGBoost (eXtreme Gradient Boosting trees) regressor. Both PixelHop and RFT are green learning tools. GIRD yields a low carbon footprint due to its smaller model sizes and lower computational complexity. Besides, its performance scales well with small training datasets. Experiments on synthetic and real circuits are given to demonstrate the superior performance of GIRD. The model size and the complexity, measured by the Floating Point Operations (FLOPs), of GIRD are only 10-3 and 10-2 of deep-learning methods, respectively.