cv

The Chinese University of Hong Kong, Shenzhen

Bachelor of Science in Computer Science and Engineering, School of Data Science

Nanyang Technological University

Exchange Program to College of Computing and Data Science

Huawei

AI Engineer Intern

• Applied the quantization technique RaBitQ to BGE-Reranker. Use VNNI AVX512 and oneDNN engine to accelerate matrix multiplication block in forward inference.
• Benchmark the MemOS with and without vector database and graph database. Integrated Membench datasets into the framework. Wrote comprehensive report on the performance of MemOS, mem0 on locomo, Longmemeval, Membench datasets.
• Research on qwen omni vision lanaguage models, and the multi-vector indexing of maxsim and fast-plaid appeared in colbert and colpali.

SenseTime

Model Evaluation Intern

• Assisted in internal evaluations of the Sensechat and Cantonese multimodal large language model series, automating competitive end-to-end model evaluations via Python scripts.
• Enhanced SenseBench for model benchmarking by integrating GPT-turbo as a judge for automated scoring, refining prompts for specific ToB business cases to improve scoring accuracy.
• Evaluated model performance on test sets, provided feedback based on RLHF.
• Created and maintained subjective test datasets.

FXKN, CUHK (Shenzhen)

Wechat Miniprogram Developer

• Developed a WeChat Mini Program using WXML, WXSS, and WXS within the WeChat Developer Tools framework. Implemented basic features that direct users to result pages based on abnormal health indicators in their medical reports, providing potential disease risk assessments.