Atlas / People / Detail

Zuquan Song

Zuquan Song is publicly listed as a ByteDance Seed author on Seed technical reports and on ByteDance Seed's OSDI 2025 publication "Understanding Stragglers in Large Model Training Using What-if Analysis." dblp also attributes 2024-2025 publications on LLM training infrastructure, checkpointing, faulty-machine detection, and GPU communication overlap to Zuquan Song.

ByteDance Seed researcher focused on large-model training infrastructure and distributed systems.1 organizations2 reports

Profile status: updated

updated 4 public sources

LLM training systemstraining infrastructuredistributed systemscheckpointingGPU communication

Current frame

ByteDance Seed researcher focused on large-model training infrastructure and distributed systems.

Extended note

Public sources link Zuquan Song to ByteDance Seed research and engineering work around large-model training systems. ByteDance Seed's publication page lists Song as an author of the OSDI 2025 paper "Understanding Stragglers in Large Model Training Using What-if Analysis," and the provided Seed technical reports also list Song as an author. dblp's bibliography for Zuquan Song shows a 2024-2025 publication record including work on LLM training infrastructure, unified checkpointing for foundation models, faulty-machine detection in distributed model training, and GPU communication overlap. No strong public source was found for education history, a personal homepage, or a more specific current title, so those fields are left empty.