@MaxForAI: Breaking — ByteDance Seed pre-training lead Quanquan Gu departs. Quanquan Gu, Associate Professor at UCLA Computer Science, earned his bachelor's and master's degrees from Tsinghua University in Automation/Control, and his PhD in Computer Science from UIUC in 2014 under the supervision of Jiawei Han.

X AI KOLs Timeline News
byte-dance resignation ai-pretraining ai-for-science people-move seed-team

Summary

Quanquan Gu, head of ByteDance Seed pre-training, has announced his departure. He previously led teams for AI for Science (e.g., SeedFold, SeedProteo) and cutting-edge LLM pre-training, contributing to Seed2.0 training.

Breaking — ByteDance Seed pre-training lead Quanquan Gu has departed. Quanquan Gu Associate Professor, UCLA Computer Science Bachelor's and Master's degrees from Tsinghua University in Automation/Control PhD in Computer Science from UIUC in 2014 Advisor: Jiawei Han He joined ByteDance Seed in 2023. Many may not know exactly what he worked on. Over the past three years, he was involved in two core tracks at Seed: AI for Science and cutting-edge LLMs. On the AI for Science side, he led the development of SeedFold, SeedProteo, and the DPLM series. SeedFold is a biomolecular structure prediction model from ByteDance Seed, claimed in its paper to surpass AlphaFold3 on multiple protein-related tasks in FoldBench; SeedProteo is a de novo all-atom protein design model for protein binder design. In 2025, he also joined the LLM pre-training effort, forming an LLM optimization and scaling team, and participated in training Seed2.0. He was one of the most central people at ByteDance Seed over the past few years, bridging bio-AI, foundation models, and scaling capabilities. He understands model training, can apply capabilities to specific scientific problems, and has led high-intensity projects. He's going to be heavily recruited! Best wishes to Dr. Gu.
Original Article
View Cached Full Text

Cached at: 06/02/26, 03:42 PM

Breaking: ByteDance Seed’s Pre-training Lead Quanquan Gu Resigns.

Quanquan Gu Associate Professor, UCLA Computer Science Department Bachelor’s and Master’s in Automation/Control from Tsinghua University Ph.D. in Computer Science from UIUC in 2014 Advisor: Professor Jiawei Han

He joined ByteDance Seed in 2023. Many may not know exactly what he worked on.

Over the past 3 years, he was involved in two core tracks at Seed simultaneously: AI for Science and cutting-edge LLMs.

On the AI for Science side, he led the development of SeedFold, SeedProteo, and the DPLM series.

SeedFold is ByteDance Seed’s biomolecular structure prediction model, which the paper claims surpasses AlphaFold3 on multiple protein-related tasks in FoldBench. SeedProteo is a de novo all-atom protein design model for protein binder design.

In 2025, he also joined the LLM pre-training effort, established the LLM optimization and scaling team, and participated in the training of Seed 2.0.

Over the past few years at ByteDance Seed, he has been one of the most central figures bridging biological AI, foundation models, and scaling capabilities.

He understands model training, can apply capabilities to concrete scientific problems, and has truly led high-intensity projects.

He’s going to be heavily recruited!

Wishing Professor Gu all the best.

Quanquan Gu (@QuanquanGu): Today marks my last day at ByteDance Seed.

Over the past 3 years, I had the opportunity to work across two of the most exciting frontiers in AI: AI for Drug Discovery and building frontier LLMs. Few opportunities in a career allow one to work simultaneously on curing disease and

Similar Articles

@0xLogicrw: Weijie Su, a 2007 alumnus of Peking University's School of Mathematical Sciences, a Stanford PhD in Statistics, and a full professor at the Wharton School of the University of Pennsylvania, officially announced that he has joined OpenAI during his leave from Wharton. Su said that after a decade, he is back in the Bay Area and will participate in AI

X AI KOLs Timeline

Weijie Su, a Peking University alum and full professor at Wharton, announced his joining OpenAI to participate in AI model training. He has also been promoted to full professor.

@MaxForAI: Former NVIDIA and Meta researcher, xAI head of world model Ethan He just announced his departure. Many may not know exactly what he was responsible for. He was a core member of Grok Imagine (including video generation) from the 0-to-1 stage. He said when he joined xAI, Grok Imagine had nothing...

X AI KOLs Timeline

Ethan He, former head of world model at xAI and key contributor to Grok Imagine video generation, announced his departure. He built the multimodal video model from scratch in three months after joining xAI in July 2025.

@MaxForAI: Yesterday, ByteDance Seed open-sourced a very interesting checkpoint, TaskMem. It is trained on Qwen3-VL-30B-A3B, with the goal not being to directly answer questions, but to enable multimodal Agents to learn to generate more useful long-term memory from video/environment streams. The key is to let the Agent learn in continuous video…

X AI KOLs Timeline

ByteDance Seed has open-sourced the TaskMem checkpoint, trained on Qwen3-VL-30B-A3B. It uses two-stage reinforcement learning to enable multimodal Agents to learn to generate long-term memory from video streams, achieving significant improvements on benchmarks such as VideoMME and EgoLife.