Daniel Han

Daniel Han, co-founder and CEO of Unsloth, is a hands-on ML systems builder who has made LLM fine-tuning and RL training dramatically faster and more memory-efficient. His open-source Unsloth library achieves up to 2x faster training with up to 70% less VRAM, has amassed 67k+ GitHub stars, and has driven 250M+ model downloads (making Unsloth the 10th most-followed org on Hugging Face). Attend his session for deep, implementable insights on squeezing performance from open models via custom Triton kernels, quantization, and RL techniques.