Xin Qiu
Software Engineer at Intel Corporation
Minhang District, Shanghai, China
Summary
🤩
Rockstar🎓
Top SchoolXin Qiu is a software engineer with a decade of experience at Intel Asia Pacific R&D, combining deep systems knowledge with practical ML deployment skills. Based in Shanghai, he holds a master's in computer science from Wuhan University and has worked across distributed systems, high-availability components, and big-data deployments since his early Intel internship. Recently he has focused on accelerating local LLM inference on Intel XPU hardware, contributing notable work to intel/ipex-llm—especially converting GPTQ quantized models to GGML and adding INT4/INT5/INT8 support and fused RMSNorm optimizations. His contributions bridge model-level quantization techniques and device-specific performance tuning, making large models run faster and more compatibly on diverse Intel accelerators. Known for shipping practical tooling and conversion scripts (including safetensors support), he brings a pragmatic, performance-first approach to ML engineering and production readiness.
10 years of coding experience
Master's degree, Computer Science, Master's degree, Computer Science at Wuhan University
Github Skills (11)
transformers10
quantization10
pytorch10
quants10
machine-learning10
gm10
gml10
python10
model-optimization10
xpu10
cuda4
Programming languages (6)
Github contributions (5)
intel/ipex-llm
Sep 2016 - Jan 2023
Accelerate local LLM inference and finetuning (LLaMA, Mistral, ChatGLM, Qwen, DeepSeek, Mixtral, Gemma, Phi, MiniCPM, Qwen-VL, MiniCPM-V, etc.) on Intel XPU (e.g., local PC with iGPU and NPU, discrete GPU such as Arc, Flex and Max); seamlessly integrate with llama.cpp, Ollama, HuggingFace, LangChain, LlamaIndex, vLLM, DeepSpeed, Axolotl, etc.
Role in this project:
ML Engineer
ML EngineerContributions:2 releases, 236 reviews, 1256 commits in 6 years 4 months
Contributions summary:Xin primarily contributed to the conversion of GPTQ quantized models to GGML format, enabling faster inference and wider compatibility. They added and improved scripts for model conversion, specifically supporting safetensors models. The user also enhanced the library by incorporating INT4, INT5, and INT8 quantization, along with related optimizations such as fused RMSNorm and other performance enhancements for XPU devices, demonstrating a focus on model optimization and deployment.
llm-inferencellama2pythonfinetuningllama
qiuxin2012/analytics-zoo
Oct 2017 - Oct 2021
Deep learning powered big data analytics using BigDL on Apache Spark
Contributions:24 PRs, 630 pushes, 159 branches in 4 years
bigdlanalyticsdata-analyticsdata-sciencedeep-learning
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.