2026. Research Scientist at the Efficient AI Team, NVIDIA Research.
Haozhe Liu
Research Scientist, Efficient AI, NVIDIA Research
I am a Research Scientist at NVIDIA Research, working with Enze Xie and Song Han, and a founding member of its Singapore lab. My research is on recursive self-improvement: agent harnesses that discover their own improvements, and agentic workflows that make foundation models faster and more efficient.
I received my Ph.D. from KAUST, advised by Jürgen Schmidhuber. During my Ph.D., I was a Research Scientist Intern at Meta, working on text-to-image and video generation models. My work includes 40+ papers with 1.9K+ citations, a NeurIPS 2023 workshop Best Paper Award, open-source projects with 10K+ GitHub stars, and models with 1M+ downloads on Hugging Face.
Google Scholar · GitHub · LinkedIn · Email
News
2026. Defended PhD dissertation at KAUST.
2024. Two-year research experience at Meta AI.
2021. One-year research experience at Tencent.