cv

Research Intern

AntGroup AI Infrastructure Group

Build efficient training system over heterogeneous GPUs.

• (co-)Design quick checkpoint scheme for Large Language Models training.
• Optimize distributed parallelism for Parameter Efficient Fine-tuning (PEFT).

PhD

University of Viriginia, Charlottesville, VA, United States

Computer Science

Bachelor

Sichuan University, Chengdu, China

Computer Engineering

DLRover

DLRover is an automatic system aiming to train large AI models easy, stable, fast and green.

• Deep Learning
• AutoML
• Distributed Computing