CV
Summary
AI Infra Engineer at ByteDance. M.S. in Software Engineering from Zhejiang University. Focus on LLM/DiT inference optimization and speculative decoding.
Education
- Software Engineering2026-03Zhejiang University
- Computer Science and Technology2023-06Zhejiang University of TechnologyGPA: 4.23/5
Work Experience
- AI Search ML Infra Engineer2026-01 -ByteDanceMulti-modal LLM speculative decoding training and RL post-training async pipeline refactoring.
- Large Model Inference Optimization Intern2025-06 - 2025-12Kuaishou - Keling Technology DepartmentDiT model feature cache reuse optimization. Work accepted to ICLR 2026.
- Navigation Group Algorithm Intern2024-09 - 2025-04Hangzhou Feibu TechnologyEnd-to-end predictive-interactive modeling and RL optimization.
Skills
Deep Learning & Inference Optimization
- LLM Inference
- DiT Inference
- Speculative Decoding
- EAGLE3
- vLLM
- xDiT
- CUDA
- FP8
- Tensor Parallelism
- CUDA Graph
Publications
- SCALINGCACHE: Extreme Acceleration of DiTs Through Difference Scaling and Dynamic Interval Caching2026ICLR 2026DiT model acceleration through feature caching, achieving 2.5x-3.0x end-to-end speedup.
Languages
- ChineseNative
- EnglishProfessional