Michael Wyatt

Senior Software Engineer

Michael Wyatt is a senior software engineer at Snowflake, where he works on the AI team. His focus is on optimizing both training and inference for large language models through ArcticTraining and ArcticInference. Michael earned his PhD in Computer Science from the University of Delaware in 2020. Prior to Snowflake, he contributed to the LBANN project at Lawrence Livermore National Laboratory and helped advance large-scale model training as part of the DeepSpeed team at Microsoft. His work spans performance engineering, system-level optimization, and efficient deployment of cutting-edge AI models.

Arctic Long Sequence Training (ALST): Scalable And Efficient Training For Multi-Million Token Sequences

Snowflake's ALST enables scalable training of long-context models with up to 15 million tokens using Hugging Face and DeepSpeed, all without custom modeling code.