Pengfei He
Research and build full-stack deep learning solutions for real world problems.

About Me

Hi! My name is Pengfei He (pronounced as "Ho"), and I also go by Murphy. I obtained my bachelor of science with honors degree major in ACMS and minor in Linguistics. I have four years of both industrial and academic work experiences, and I am currently looking for a full-time machine learning engineer position in the field of Natural Language Processing.

Regarding research, I am also a part-time NLP researcher at UW NLP labs. My research interests lie broadly in Natural Language Processing and Machine Learning, and currently I am interested in the following fields:

Algorithm

• Instruction Tuning (technical report)
• Parameter Efficient Fine Tuning in Efficient AI (technical report)
• Reinforcement Learning with Human/AI Feedback (RAINIER)

System

• High Performance Training in a Distributed Environment
  • Blogs:
    • Data Parallelism
    • Pipeline Parallelism
    • Tensor Parallelism
    • Hybrid Parallelism
    • GPU communication
    • Training Memory Estimation
  • Projects:
    • Large-Scale PEFT Experiments on clusters
• Building, Testing, Maintaining and Serving Home Cluster with SOTA Home Workstation Performance

Application

• Business Application of AI Agent (Founded AdaBit AI, actively developing)
• Bring NLP algorithms to Biological Language Processing (still exploring potential roles/projects)