About
Chao Wang is a Staff Engineer on the LongCat team at Meituan (Beidou Talent Program). The work focuses on LLM inference, including reliability, performance, and cost efficiency. Chao Wang received a Ph.D. in Computer Science and Engineering from The Chinese University of Hong Kong in 2026, advised by Prof. Ming-Chang Yang. Chao Wang’s publications include papers at OSDI, ICML, SC, and VLDB.
Education
PhD in Computer Science and Engineering
The Chinese University of Hong Kong (CUHK): 2020 - 2026
Bachelor of Engineering in Computer Science and Technology
Huazhong University of Science and Technology (HUST): 2016 - 2020
Publications
Timnah: A Tier-Memory-Native Hashing for Modern Tiered Memory Systems
VLDB'27
Prefill-Decode Aggregation or Disaggregation? Unifying Both for Goodput-Optimized LLM Serving
SC'26, FAISys Talk, ArXiv, 公众号
TileSparse: Arithmetic-Intensity-Aware Sparse Attention for Compute-Bound LLM Decoding
SEPH: Scalable, Efficient, and Predictable Hashing on Persistent Memory
OSDI'23 (First OSDI paper from CUHK)
SparseServe: Unlocking Parallelism for Dynamic Sparse Attention in Long-Context LLM Serving
CARINA: An Efficient CXL-Oriented Embedding Serving System for Recommendation Models
Aceso: Achieving Efficient Fault Tolerance in Memory-Disaggregated Key-Value Stores
SOSP'24 (Second SOSP paper from CUHK)
Academic Service
External Reviewer
OSDI'26, Eurosys'26, ICML'26, SoCC'26, HPCA'25, NSDI'25, FAST'24, ATC'24, SIGMOD'24, TOS'24, DAC'20-21, NAS'24, ASP-DAC'24, CODES'20~23, ICCAD'22, RTAS'21