Tri Dao / Princeton & Together AI — AI Researcher

**Tri Dao** is an Assistant Professor at **Princeton University** and the Chief Scientist of **Together AI**. He is the creator of **FlashAttention** — arguably the single most impactful systems-level contribution to modern AI. FlashAttention is a GPU-aware algorithm that computes exact attention 2-4x faster while using 5-20x less memory than the standard implementation, and it is now used in **every major LLM training and inference pipeline in the world** — PyTorch, Hugging Face, vLLM, and every frontier lab's internal stack. It is rare for one algorithm to so completely and universally transform the field.