Mingjun Wang

Ph.D. Student · Department of Computer Science and Engineering · The Chinese University of Hong Kong

LLM4EDA / AI4EDA 3D IC

I develop learning methods and design flows for integrated circuits, connecting circuit representation learning and LLM-based automation with 3D physical implementation. My work spans early-stage timing prediction, RTL-to-GDS flows, and efficient fault simulation.

CV (PDF)

About

I am a Ph.D. student in Computer Science and Engineering at The Chinese University of Hong Kong, advised by Prof. Bei Yu. My research has two closely connected directions: LLM4EDA / AI4EDA and 3D IC design automation.

On the AI side, I study multi-modal circuit representations, cross-stage knowledge distillation, and LLM-based verification and optimization. On the 3D IC side, I develop legality-aware RTL-to-GDS flows and early-stage timing models for hybrid-bonded designs. My current interests include tool-using agents for chip implementation and the interaction of data paths, bonding parasitics, and clock networks in setup/hold timing closure.

Before joining CUHK, I received my master's degree in Electronic Information from the University of Chinese Academy of Sciences and the Institute of Computing Technology, Chinese Academy of Sciences, advised by Prof. Xiaowei Li and Prof. Huawei Li at the State Key Laboratory of Processors. I received my bachelor's degree in Microelectronics Science and Technology from Peking University, and have industrial EDA experience with CASTEST on fault simulation and ISO 26262-oriented functional safety analysis.

News

2026/11 · ASP-DAC 2027: CU-3D is an invited paper; HOIST and Logic Resynthesis for Cross-Die Dependencies are accepted. I am a co-corresponding author of HOIST.

2026/11 · MigSizer is accepted at ICCD 2026 for physical-aware analog/RF circuit migration.

2026/07 · ICCAD 2026: Distill3D, LibPilot, and ChatFCM accepted.

2026/05 · IEEE TCAD Early Access: "Bit-Compressed Concurrent Fault Simulation with Efficient Fault List" is online.

2026/05 · ICML 2026: AnalogVerifier accepted.

2026/04 · ICLR 2026: CircuitNet 3.0 accepted as a poster paper.

2026/04 · DAC 2026: ATLAS accepted for FIT prediction with asynchronous topological learning.

2025/12 · 2nd Place at EDAthon 2025, Programming Competition on Electronic Design Automation.

Research Interests

LLM4EDA / AI4EDA

Circuit representation learning, knowledge distillation, and LLM-based design. My current interests include domain-adaptive continued pretraining, SFT, and tool-feedback reinforcement learning, grounded in functional and physical verification.

3D IC Design Automation

Legality-aware RTL-to-GDS flows, hybrid-bonding integration, and cross-stage timing prediction. I connect physical implementation with post-route analysis and setup/hold timing closure.

Fault Simulation and Functional Safety

A complementary foundation in hardware-aware simulation, parallel algorithms, and ISO 26262-oriented analysis, developed through DomSim, EPICS, ETPG, and industrial EDA work.

Education

CUHK logo
The Chinese University of Hong Kong2025 — Present

Ph.D. in Computer Science and Engineering

Advisor: Prof. Bei Yu

Research focus: LLM4EDA / AI4EDA and 3D IC design automation

ICT CAS logo
University of Chinese Academy of Sciences & ICT, CAS2022 — 2025

M.Eng. in Electronic Information · GPA 3.77/4.0 (Top 5%)

Advisors: Prof. Xiaowei Li and Prof. Huawei Li · State Key Laboratory of Processors

Peking University logo
Peking University2018 — 2022

B.Eng. in Microelectronics Science and Technology

Selected Publications

† Equal contribution. ‡ Co-corresponding authors.

2027
ASP-DAC 2027ConferenceInvited Paper

CU-3D: A Scalable Legality-Aware RTL-to-GDS Flow for 3D-IC Physical Design

Mingjun Wang, Guanqi Li, Yihan Wen, Lancheng Zou, Mingjun Li, Feng Gu, Fengrui Liu, Jianan Mu, Yuxuan Zhao, Tsung-Yi Ho, Huawei Li, Sung Kyu Lim, Bei Yu.

IEEE/ACM Asia and South Pacific Design Automation Conference · Tokyo, Jan. 25–28, 2027.

A scalable RTL-to-GDS flow that integrates physical legality into 3D IC implementation.

ASP-DAC 2027ConferenceAcceptedCo-corresponding ‡

HOIST: A High-Order Interaction Surrogate for Sample-Efficient High-Sigma Yield Estimation in High-Dimensional Circuits

Zhongru Xiong, Mingjun Wang‡, Yanfang Liu, Tsung-Yi Ho, Bei Yu, Binwu Zhu‡.

IEEE/ACM Asia and South Pacific Design Automation Conference · Tokyo, Jan. 25–28, 2027.

Models high-order interactions to improve sample efficiency in high-sigma circuit yield estimation.

ASP-DAC 2027ConferenceAccepted

Logic Resynthesis for Cross-Die Dependencies in Hybrid Bonded 3D ICs

Fangzhou Liu, Yuhao Ji, Lancheng Zou, Jiawei Liu, Mingjun Wang, Gengjie Chen, Guannan Guo, Bei Yu.

IEEE/ACM Asia and South Pacific Design Automation Conference · Tokyo, Jan. 25–28, 2027.

Logic resynthesis addressing dependencies across dies in hybrid-bonded 3D ICs.

2026
ICCAD
ICCAD 2026ConferenceAccepted

Distill3D: Cross-Stage Knowledge Distillation for Early-Stage 3D IC Timing Prediction

Mingjun Wang†, Lancheng Zou†, Guanqi Li†, Feng Gu, Yuntao Lu, Tianmeng Yang, Boyu Han, Jianan Mu, Huawei Li, Sung Kyu Lim, Bei Yu.

Conference paper · International Conference on Computer-Aided Design, 2026.

Cross-stage knowledge distillation for accurate early-stage timing prediction in 3D IC design flows.

ICLR
ICLR 2026 PosterConference

CircuitNet 3.0: A Multi-Modal Dataset with Task-Oriented Augmentation for AI-Driven Circuit Design

Mingjun Wang†, Yihan Wen†, Yuntao Lu, Fengrui Liu, Yuxiang Zhao, Boyu Han, Jianan Mu, Yibo Lin, Runsheng Wang, Bei Yu, Huawei Li.

Conference paper · International Conference on Learning Representations, 2026.

More than 15,000 circuit instances from 8,659 validated designs, with multi-stage data for timing and power prediction.

IEEE TCAD 2026Journal

DomSim: Hardware-Aware Hybrid Fault Simulation With Dominator Tree-Guided Partitioning

Mingjun Wang†, Hui Wang†, Feng Gu, Zizhen Liu, Jianan Mu, Shengwen Liang, Zhongkai Yu, Zheng Liang, Jun Gao, Jiaping Tang, Jing Ye, Xiaowei Li, Bei Yu, Huawei Li.

IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 45(6): 2725–2738, 2026. Early Access: 2025.

Hardware-aware hybrid fault simulation with dominator-tree partitioning; 10.29× average speedup over the commercial baseline in the paper's experiments.

DAC
DAC 2026ConferenceCo-first †

ATLAS: Asynchronous Topological Learning for Accurate FIT Prediction via Decoupled Graph Neural Networks

Yicheng Liu†, Mingjun Wang†, Songwei Pei, Yuntao Lu, Zizhen Liu, Boyu Han, Huawei Li, Shangguang Wang.

Conference paper · Design Automation Conference, 2026.

A decoupled graph learning approach for accurate FIT prediction.

ICML
ICML 2026ConferenceCo-first †

AnalogVerifier: A Neuro-Symbolic Framework for Analog Circuit Verification

Yanfang Liu†, Mingjun Wang†, Peng Xu, Rongliang Fu, Bei Yu, Tsung-Yi Ho.

Conference paper · International Conference on Machine Learning, 2026.

A neuro-symbolic verification framework for analog circuits.

ICCAD
ICCAD 2026ConferenceAccepted

LibPilot: Knowledge-Constrained Dual-Agent Reasoning for Standard Cell Library Tuning

Yanfang Liu, Zijin Cheng, Mingjun Wang, Rongliang Fu, Chao Wang, Bei Yu.

Conference paper · International Conference on Computer-Aided Design, 2026.

Knowledge-constrained dual-agent reasoning for efficient standard-cell library tuning.

ICCAD
ICCAD 2026ConferenceAccepted

ChatFCM: Comprehension-Synthesis Decoupling for LLM-Based Functional Coverage Model Generation

Yuan Pu, Yudong He, Zhenghao Chen, Mingjun Wang, Bingkun Yao, Zhaotan Lin, Qin Chen, Bing Li, Song Chen, Zhuolun He, Wenjian Yu, Yanfeng Li, Bei Yu.

Conference paper · International Conference on Computer-Aided Design, 2026.

A comprehension-synthesis decoupled framework for LLM-based functional coverage model generation.

ICCD 2026ConferenceAccepted

MigSizer: Physical-Aware Sizing Improving Analog/RF Circuit Migration

Yanfang Liu, Peng Xu, Yapeng Li, Mingjun Wang, Leilei Jin, Rongliang Fu, Tinghuan Chen, Wei Xing, Bei Yu, Tsung-Yi Ho.

IEEE International Conference on Computer Design · Hong Kong, Nov. 16–18, 2026.

Physical-aware sizing for analog/RF circuit migration.

TCAD
IEEE TCAD Early AccessJournal

Bit-Compressed Concurrent Fault Simulation with Efficient Fault List

Feng Gu, Zhiteng Chao, Hui Wang, Jianan Mu, Mingjun Wang, Jun Gao, Yonghao Wang, Jing Ye, Xiaowei Li, Huawei Li.

Journal article · IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, Early Access, 2026.

A bit-compressed concurrent fault simulation method with efficient fault-list organization for design-for-test workflows.

ASP-DAC
ASP-DAC 2026ConferenceInvited Paper

LLM-Assisted Circuit Verification: A Comprehensive Survey

Hongduo Liu, Yuntao Lu, Mingjun Wang, Xufeng Yao, Bei Yu.

Conference paper · Asia and South Pacific Design Automation Conference, 2026.

A survey of LLM-assisted methods for circuit verification.

2025
ICML
ICML 2025ConferenceSpotlight

Bridging Layout and RTL: Knowledge Distillation based Timing Prediction

Mingjun Wang†, Yihan Wen†, Bin Sun, Jianan Mu, Juan Li, Xiaoyi Wang, Jing Justin Ye, Bei Yu, Huawei Li.

Conference paper · International Conference on Machine Learning, 2025.

Transfers physical timing knowledge from layout-aware teachers to efficient RTL-level student models.

DAC
DAC 2025Conference

MOSS: Multi-Modal Representation Learning on Sequential Circuits

Mingjun Wang, Bin Sun, Jianan Mu, Feng Gu, Boyu Han, Tianmeng Yang, Xinyu Zhang, Silin Liu, Yihan Wen, Hui Wang, Jun Gao, Zhiteng Chao, Husheng Han, Zizhen Liu, Shengwen Liang, Jing Ye, Bei Yu, Xiaowei Li, Huawei Li.

Conference paper · Design Automation Conference, 2025.

Multi-modal representation learning for sequential circuits.

DAC
DAC 2025Conference

EPICS: Efficient Parallel Pattern Fault Simulation for Sequential Circuits via Strongly Connected Components

Mingjun Wang, Hui Wang, Jianan Mu, Xinyu Zhang, Bin Sun, Yihan Wen, Zizhen Liu, Feng Gu, Jun Gao, Shengwen Liang, Jing Ye, Xiaowei Li, Huawei Li.

Conference paper · Design Automation Conference, 2025.

Parallel pattern fault simulation for sequential circuits through SCC-aware decomposition.

ASP-DAC
ASP-DAC 2025ConferenceBest Paper Nomination

ETPG: Efficient Transition Fault Simulation via Dual-Strategy Pattern Parallelism and Gate Restructuring

Mingjun Wang, Hui Wang, Zizhen Liu, Feng Gu, Jianan Mu, Jiaping Tang, Jun Gao, Huawei Li, Jing Ye, Xiaowei Li.

Conference paper · Asia and South Pacific Design Automation Conference, 2025.

Dual-strategy pattern parallelism and gate restructuring for efficient transition fault simulation.

ICCAD
ICCAD 2025ConferenceCo-first †

VIRTUAL: Vector-based Dynamic Power Estimation via Decoupled Multi-Modality Learning

Yuntao Lu†, Mingjun Wang†, Yihan Wen, Boyu Han, Jianan Mu, Huawei Li, Bei Yu.

Conference paper · International Conference on Computer-Aided Design, 2025.

Decoupled multi-modality learning for vector-based dynamic power estimation.

ISEDA
ISEDA 2025Conference

DFTS: An Efficient Design-for-Test Flow for Scan Design

Jun Gao, Mingjun Wang, Jiangwang Liu, Wenjie Li, Zizhen Liu, Jing Ye, Huawei Li.

Conference paper · International Symposium of EDA, 2025.

An efficient design-for-test flow for scan design.

2024
ITC-Asia
ITC-Asia 2024Conference

Efficient Functional Safety Method for Gate-Level Fine-Grained Digital Circuits with ISO-26262

Mingjun Wang, Hui Wang, Jianan Mu, Zizhen Liu, Jun Gao, Jing Ye, Huawei Li, Xiaowei Li.

Conference paper · International Test Conference in Asia, 2024.

A gate-level fine-grained functional safety method for ISO 26262-oriented digital circuits.

ICCAD
ICCAD 2024ConferenceBest Paper Nomination

DDP-Fsim: Efficient and Scalable Fault Simulation for Deterministic Patterns with Two-Dimensional Parallelism

Feng Gu, Mingjun Wang, Jianan Mu, Zizhen Liu, Jiaping Tang, Hui Wang, Yonghao Wang, Jing Ye, Huawei Li, Xiaowei Li.

Conference paper · International Conference on Computer-Aided Design, 2024.

Efficient and scalable deterministic-pattern fault simulation with two-dimensional parallelism.

Research Experience

May 2024 — Present

LLM4EDA / AI4EDA and 3D IC Design Automation

  • Circuit learning: developed multi-modal representations and cross-stage timing/power models through MOSS, RTLDistil, CircuitNet 3.0, and Distill3D.
  • LLM-based EDA: designed and implemented functional coverage generation methods and tool-based validation in an industry-academic project; contributed to ChatFCM and knowledge-constrained library tuning with LibPilot.
  • 3D implementation: developed a research RTL-to-GDS flow integrating hybrid-bonding structures, clock-tree synthesis, routing, parasitic extraction, and timing analysis.
  • Current focus: tool-using agents that improve chip designs under functional and physical constraints, evaluated through post-route results and tool-use costs.
Jul. 2023 — Mar. 2025

Fault Simulation Research · State Key Laboratory of Processors, ICT, CAS

  • Developed hardware-aware parallel simulation using dominator-tree partitioning, strongly connected components, and pattern parallelism.
  • DomSim achieved a 10.29× average speedup over the commercial baseline in its published experiments; EPICS achieved 5.94× while preserving fault coverage in its experiments.
Sep. 2021 — Jan. 2025

Industrial Experience · CASTEST

  • Developed core fault simulation algorithms and contributed to tool integration, multi-core optimization, and industrial requirements validation.
  • Integrated parallel simulation into ISO 26262-oriented functional safety analysis workflows for automotive chips.
Sep. 2020 — Jun. 2022

RRAM-Based True Random Number Generator · PKU Undergraduate Thesis

  • Designed and implemented an RRAM-based random number generator.
  • Evaluated randomness using the NIST statistical test suite.

Honors & Awards

AwardVenue / OrganizationYear
Winner — IEEE ICLAD-DAC GenAI Hackathon, Design Verification SLM TrackIEEE ICLAD & DAC2025
2nd Place — EDAthon Programming CompetitionIEEE CEDA2025
2nd Prize (Enterprise Track) — IC Innovation Contest8th China Graduate IC Innovation Contest2025
3rd Prize (Team Competition) — IC Innovation Contest8th China Graduate IC Innovation Contest2025
International Exchange Scholarship for EmpowermentICT, CAS2025
Director's Special Award (1st / 400)ICT, CAS2024
National Scholarship (Top 2%)MoE, China2024
Special Prize — LLM4EDA Hackathon (1st / 100)National Key Laboratory of Integrated Chips and Systems2024
Academic Endeavor AwardEDA Elite Challenge2024
1st Prize Academic Scholarship (Top 15%)University of Chinese Academy of Sciences2024
Merit StudentUniversity of Chinese Academy of Sciences2023, 2024
1st Prize — EDA² Integrated Circuit Elite Challenge (1st / 230 teams)Huawei HiSilicon2023
Future Star — MOE–Huawei Intelligent Base ProgramMoE & Huawei2023
Cambricon Outstanding Student Award (10 / 300)Cambricon2023
Outstanding Contribution Award — Beijing 2022 Winter Olympics (Top 10%)Peking University2022
Peking University Social Work AwardPeking University2019, 2020

Leadership & Service

Skills

Languages
PythonC++SystemVerilogVerilogTcl
EDA Tools
Cadence InnovusSynopsys PrimeTimeSynopsys Design CompilerSynopsys VCSCadence Xcelium
AI / ML
PyTorchCUDAPyTorch GeometricTensorFlow
Development
LinuxGitLaTeX

Collaboration & Contact

I welcome collaborations on LLM4EDA / AI4EDA and 3D IC design automation, especially tool-using agents, circuit learning, RTL-to-GDS flows, and timing closure. I also work on fault simulation and hardware verification.

Email: wangmingjun000613@gmail.com · mjwang25@cse.cuhk.edu.hk

CUHK · Department of Computer Science and Engineering · Advisor: Prof. Bei Yu

Google Scholar · GitHub · CV