Chenyu Wang
I am a researcher at Harvard University. I work on AI and systems, and am envisioning Autonomous AI agents for system engineering.
This semester, I'm Course Staff for CS249r: Agentic AI for Computer Systems Design with Professor Vijay Janapa Reddi.
If you want to find my CV, it's here.
I curated an AI for Hardware/System papers collection, a small contribution to the community. Feel free to explore and share!
Research Interests
Recent Works
SLM-MUX: Orchestrating Small Language Models for Reasoning
A multi-model architecture that effectively coordinates multiple small language models, achieving up to 13.4% improvement on MATH and outperforming Qwen 2.5 72B with just two SLMs.
EPIM: Efficient Processing-In-Memory Accelerators based on Epitome
Developed a novel processing-in-memory architecture for efficient neural network acceleration, demonstrating significant improvements in energy efficiency and throughput.
Evaluating Zero-Shot Long-Context LLM Compression
Comprehensive evaluation of compression techniques for large language models with extended context windows, addressing practical deployment challenges.
Gibbon: Efficient co-exploration of NN model and processing-in-memory architecture
Designed an efficient co-exploration framework for jointly optimizing neural network models and processing-in-memory architectures, achieving state-of-the-art performance.