Xuhui Ren
PhD Candidate
Brisbane, Austrilia
Summary
🤩
RockstarXuhui Ren is a PhD candidate at the University of Queensland's DKE group with five years of hands-on experience optimizing machine learning models for production. Based in Brisbane, he contributes to Intel's widely used neural-compressor project, focusing on low-bit LLM quantization (INT8/FP8/INT4/FP4/NF4), sparsity, and performance benchmarking across TensorFlow, PyTorch, and ONNX Runtime. His work spans post-training quantization, quantization-aware training, API migrations, and improving usability through documentation and example modernization. Combining academic research with practical ML engineering, he brings a pragmatic approach to squeezing efficiency out of large models while maintaining developer-friendly tooling.
5 years of coding experience
Github Skills (9)
quantization10
pytorch10
machine-learning10
bert10
model-optimization10
python9
nlp9
auto-tuning9
sparse8
Programming languages (4)
Github contributions (5)
intel/neural-compressor
Sep 2022 - Jan 2023
SOTA low-bit LLM quantization (INT8/FP8/INT4/FP4/NF4) & sparsity; leading model compression techniques on TensorFlow, PyTorch, and ONNX Runtime
Role in this project:
ML Engineer
ML EngineerContributions:11 reviews, 11 commits, 10 PRs in 4 months
Contributions summary:Xuhui's contributions primarily revolve around optimizing and refining machine learning models within the context of the Intel Neural Compressor. The commits demonstrate an involvement in adapting quantization techniques, such as dynamic and static post-training quantization, and quantization-aware training for models like BERT. The user also focused on performance benchmarking and enabling features like batch size adjustment in the quantized models. Additionally, the user has been active in migrating examples to a new API and refining the documentation, showcasing a focus on the overall project usability.
knowledge-distillationauto-tuningcompressorsparsityintel
XuhuiRen/GenAIComps
May 2024 - Oct 2024
GenAI components at micro-service level; GenAI service composer to create mega-service
Contributions:81 pushes, 11 branches in 4 months
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.