Bio
VP of Kernels at Together AI and Assistant Professor of Computer Science and Engineering at UC San Diego, focused on efficient machine learning systems and GPU performance.
Session (1)
Additional Speaker Details
Using publicly available information we constructed an analysis to help you get a feel for this speaker before deciding to attend their session.
Fu leads Together AI's GPU-kernels team and is a lead author on ThunderKittens, FlashFFTConv, and Monarch Mixer, so his session is a chance to hear from a hardware-aware-ML researcher how to make models actually fast on modern GPUs — and why he argues software-hardware co-design (not just bigger models) unlocks the next order of magnitude of performance.
Speaking style
Deeply technicalBig-pictureBattle-tested
Recent talks (1)
GitHub
@DanFu09Recent writing (3)
- Yes, AGI Can Happen – A Computational Perspective · blog · Dec 2025
- HipKittens: Fast and Furious AMD Kernels · paper · Nov 2025
- Chipmunk: Training-Free Acceleration of Diffusion Transformers with Dynamic Column-Sparse Deltas · paper · Jun 2025
Podcasts & interviews (1)
- Episode 5: Dan Fu, Distinguished Researcher at Together AI · Delta Institutes Podcast · May 2025