Deheng Ye （叶德珩）

Short Bio: I am a Researcher at ByteDance Seed working on foundation models.

Experience: ByteDance (2026-now), Tencent (2017-2026)

Selected Projects: SeedXxx model (P&C for now); Agentic Systems on GenMedia (ui-voyager, omni-reasoner, proact); JueWu AI Series: 1.0 (supervised learning), 2.0 (reinforcement learning), 3.0 (hierarchical RL); Code understanding & generation (as a postdoc at Microsoft 2016-2017)

Selected Publications: "Adaptvision: efficient vision-language models" (2026); "More agents is all you need" (2024); "RLTF" (2023); "Mastering complex control with deep reinforcement learning" (2020); "Convolutional and recurrent neural networks for text categorization" (2017)

Adjunct Appointments: Director of the Tencent-NTU Joint Research Laboratory (CENTURY) (until 2026)

Work Performance: was promoted to become the youngest Leader, Director, and Research Partner of Tencent AI before 2025.

Others: looking for top talents via ydh@bytedance.com or ydyl1991@gmail.com

Visits Today: …