CV

Ph.D. in Computer Science

Carnegie Mellon University, United States

• Advised by Prof. Tianqi Chen and Prof. Todd C. Mowry.
• Member of Catalyst Group.

B.Eng. in Computer Science

Shanghai Jiao Tong University, China

• Member of ACM Honors Class, a pilot CS program for top talented students.
• Advised by Prof. Yong Yu.

Research Intern

OctoML

• Working on the Machine Learning Compilation online course, as a teaching assistant.

Research Intern

Catalyst Research Group, Carnegie Mellon University

• Worked on automatic tensor program optimization and sparse tensor computing.
• Advised by Prof. Tianqi Chen

TensorIR: An Abstraction for Tensorized Program Optimization

• At UW SAMPL Group.
• video

SparseTIR: A Unified Abstraction for Sparse Workload Representation and Optimization

• At CMU Catalyst Group and TVM Conference 2021.

MLC LLM, Web LLM and Web Stable Diffusion

• Web LLM and Web Stable Diffusion bring large language models and stable diffusion models completely to people's web browsers. Everything runs inside the browser accelerated by WebGPU with no server support.
• MLC LLM is a universal solution that allows any language model to be deployed natively on a diverse set of hardware backends and native applications, plus a productive framework for everyone to further optimize model performance for their own use cases. Everything runs locally with no server support and accelerated with local GPUs on your phone and laptop.

SparseTIR, A Tensor-Level Abstraction for Sparse Operator Optimization in Deep Learning

• In close collaboration with Zihao Ye from University of Washington.
• Code, paper and documentation.

Apache TVM, An End-to-End Machine Learning Compiler Framework

• Committer of the community.
• Author of over 40+ PRs, 13000+ lines of code.
• Reviewer of over 80+ PRs.

Mx-Compiler

• A toy compiler implemented in Java, from Mx* (a C- and Java-like language) to RISC-V assembly code.
• Implemented many effective optimizations. The generated code has performance close to GCC O2.
• More than 15k+ lines of code overall.

Distributed Hash Table

• Implemented two DHT protocols, Chord and Kademlia, in Go Language.
• Implemented an instant chat room system based on the Chord protocol.

Machine Learning Compilation

• Teaching assistant of this online course.
• Prepare and release course assignments. Answer questions in the discussion page.

Principle and Practice of Computer Algorithms

• Leader teaching assistant of this SJTU course.
• Advised students to implement a RISC-V simulator.
• Advised students to implement two Distributed Hash Table protocols, Chord and Kademlia, in Go Language.

Data Structure (Honor)

• Leader teaching assistant of this SJTU course.
• Taught advanced data structures which students usually do not learn in class.
• Prepared the course assignments, projects and programming exams.

• Shanghai Excellent Graduate Award

• National Scholarship (Top 0.2% nationwide)

• The 4th place of the 2018 ICPC Asia Singapore Regional Contests