HIT Alumnus Zishen Wan Joins Columbia, Wins DAC PhD Forum for Embodied AI Hardware Co-Design

Harbin Institute of Technology (HIT) alumnus Zishen Wan began his undergraduate studies in electrical engineering at HIT in 2014. In 2018, he received the Chiang Chen Overseas Graduate Scholarship to pursue a master's degree in electrical engineering at Harvard University under Professor Vijay Janapa Reddi , graduating in 2020.

During his master's, Wan contributed to two notable research projects: one on joint floating-point encoding and hardware design, and another on hardware-software co-design for autonomous machines such as robots. These works earned the DAC 2020 Best Paper Award and were selected as IEEE Computer Architecture Letters Best of CAL for the year.

After Harvard, Wan entered the PhD program at Georgia Tech's School of Electrical and Computer Engineering , advised by Professors Arijit Raychowdhury and Tushar Krishna . His doctoral research centered on the computing demands of embodied AI and neuro-symbolic reasoning — specifically, how robots perceive environments, estimate state, and plan actions, each with distinct speed and memory requirements. Wan's thesis proposed co-designing software systems, processor architectures, and chips together to meet these needs, developing specialized architectures and taping out chip implementations.

His PhD achievements include:

Selection as a Rising Star in Machine Learning and Systems (2023) and Rising Star in Cyber-Physical Systems (2024).

Receipt of the Baidu PhD Fellowship 2025 (one of 10 global recipients).

First place at the DAC PhD Forum 2025 for his dissertation work.

Post-graduation: ACM FCCM Outstanding Dissertation Award and Georgia Tech Colonel Oscar P. Cleaver Dissertation Award .

In October 2025, ACM SIGDA announced him as the 2026 Outstanding Dissertation Award winner in Electronic Design Automation — the third major dissertation award for the same thesis.

In 2025, Wan completed his PhD and returned to Harvard as a postdoctoral researcher, again collaborating with Vijay Janapa Reddi. In July 2025, he was honored with the WAIC Yunfan Award at the World Artificial Intelligence Conference. In August 2025, he joined Columbia University's Computer Science Department as an Assistant Professor and established Wan Lab .

Wan Lab pursues two main research thrusts:

Computing systems for embodied AI, reasoning, and agents — continuing the co-design methodology across software, architecture, and silicon.

AI agents participating in computer systems design — exploring the extent to which AI agents can perform tasks such as code generation, compiler optimization, processor design, and chip verification.

In June 2025, Wan and collaborators introduced ArchEval , a benchmark for evaluating AI agents on computer architecture design tasks. In September 2025, another lab project, Hawkeye , was accepted to NeurIPS 2026 ; it investigates using hardware characteristics to optimize GPU kernels under limited supervision. The lab's next steps are to verify whether agent-proposed optimizations yield real hardware performance gains and whether they generalize across more design tasks.