Chenyu Wang
PhD Student in Electrical and Computer Engineering
Harvard University
I work at the intersection of artificial intelligence and computer systems. My research focuses on AI agents for system and architecture design, world models, domain-specific foundation models, and efficient systems for compositional intelligence. I am currently especially interested in world models for planning, simulation, and agentic decision-making. I am advised by Prof. Vijay Janapa Reddi and Prof. Yilun Du. Before Harvard, I received my M.S.E. from Princeton University and my B.E. from Tsinghua University.
chenyu_wang [at] seas [dot] harvard [dot] edu · Resume · Google Scholar · LinkedIn
Cambridge, Massachusetts
News
Research
My long-term goal is to build AI systems that can participate in the engineering process: understanding workloads, reasoning about physical and computational constraints, using simulators and tools, and proposing designs that are both effective and auditable.
- AI for systems: agents and foundation models for architecture reasoning, design-space exploration, and hardware/software co-design.
- World models: structured representations of environments that support prediction, planning, simulation, and agentic decision-making.
- Compositional AI: coordination, specialization, and adaptation across multiple language models and agents.
- Efficient AI: algorithm–architecture co-design, processing-in-memory, and efficient inference.
Publications
* denotes equal contribution2026
-
ArchEval: Measuring AI Agents as Computer Architects
Architecture 2.0 Workshop at ISCA, 2026. [paper]
-
AgentDSE: Reasoning-Augmented Architectural Design Space Exploration
MLArchSys Workshop at ISCA, 2026. [paper]
-
Economy of Minds: Emerging Multi-Agent Intelligence with Economic Interactions
-
QuArch: A Benchmark for Evaluating LLM Reasoning in Computer Architecture
International Conference on Machine Learning (ICML), 2026. [paper] [project]
-
SLM-MUX: Orchestrating Small Language Models for Reasoning
International Conference on Learning Representations (ICLR), 2026. [paper] [project]
-
GenAI for Systems: Recurring Challenges and Design Principles from Software to Silicon
Preprint, 2026. [paper]
-
Compositional AI Beyond LLMs: System Implications of Neuro-Symbolic-Probabilistic Architectures
ACM International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS), 2026. [paper]
2024
-
EPIM: Efficient Processing-In-Memory Accelerators based on Epitome
Design Automation Conference (DAC), 2024. [paper]
-
Evaluating Zero-Shot Long-Context LLM Compression
Preprint, 2024. [paper]
2023
-
Gibbon: An Efficient Co-Exploration Framework of NN Model and Processing-In-Memory Architecture
IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems (TCAD), 2023. [paper]
-
DeepGuiser: Learning to Disguise Neural Architectures for Impeding Adversarial Transfer Attacks
Preprint, 2023. [paper]
2022
-
Gibbon: Efficient Co-Exploration of NN Model and Processing-In-Memory Architecture
Design, Automation and Test in Europe Conference (DATE), 2022. [paper]
See Google Scholar for the most current publication record.
Professional Service
- Reviewer: NeurIPS, ICML, and COLM.
- Organizing team, Architecture 2.0 Workshop on AI for Computing Systems Design, ISCA 2026.
Resources
I maintain a searchable AI for Hardware and Systems paper collection covering computer architecture, EDA, compilers, systems, and hardware design.
