Guanghua Yu
副总经理 at 上海智向信息科技有限公司
Shanghai, Shanghai, China
Summary
🤩
Rockstar🎓
Top SchoolGuanghua Yu is an experienced product and technology executive with over a decade building industrial automation and smart-home products, and he currently serves as Vice General Manager leading technical product development and commercialization in Shanghai. A Tsinghua-trained software engineer and former co-founder, he blends hands-on R&D, wireless and mobile systems expertise, and product marketing to bring experimental ideas into market-ready devices. His open-source contributions to the PaddlePaddle ecosystem—especially in model compression (PaddleSlim), object detection, and core API fixes—reflect deep practical knowledge of ML model deployment and quantization. Guanghua’s background spans founding startups, leading intelligent solutions and platforms, and improving developer-facing docs and tooling, showing he values both product UX and engineering rigor. Notably, he has moved from hardware-centric smart-home experimentation to contributing to large-scale ML frameworks, bridging embedded product constraints with modern model-compression techniques.
10 years of coding experience
11 years of employment as a software developer
Bachelor's degree, Computer Software Engineering, 2003, Bachelor's degree, Computer Software Engineering, 2003 at 清华大学学生, 高分子材料, 学生, 高分子材料 at 华南理工大学研究生, 软件工程, 研究生, 软件工程 at 同济大学
Github Skills (33)
quants10
c-language10
system-configuration10
python10
tensorrt10
machine-learning10
onnx10
ch10
model-compression10
api-documentation10
custom-configuration10
deeplearning-ai10
deep-learning10
trainings10
mobilenet10
Programming languages (3)
Github contributions (5)
PaddlePaddle/PaddleDetection
Jul 2019 - Jan 2023
Object Detection toolkit based on PaddlePaddle. It supports object detection, instance segmentation, multiple object tracking and real-time multi-person keypoint detection.
Role in this project:
ML Engineer
ML EngineerContributions:638 reviews, 367 commits, 810 PRs in 3 years 7 months
Contributions summary:Guanghua primarily contributed to the PaddleDetection project by implementing various object detection models, including FaceBoxes, Blazeface, and VGG-SSD. They also added configurations and model zoo entries for these models, as well as for deformable convolution-based models. Furthermore, the user addressed bugs, added support for saving the best model during training, and integrated multi-scale evaluation capabilities for improved performance.
multi-personopenmmlabfcosmultiple-object-trackingreal-time
PaddlePaddle/PaddleSlim
Sep 2021 - Jan 2023
PaddleSlim is an open-source library for deep model compression and architecture search.
Role in this project:ML Engineer
ML EngineerContributions:343 reviews, 136 commits, 313 PRs in 1 year 3 months
Contributions summary:Guanghua primarily contributed to the PaddleSlim library for deep model compression. Their commits focused on adding post-training quantization (PTQ) functionalities, including a PTQ demo and API documentation, as well as implementing a data-free PTQ method. They also worked on integrating and supporting new features for onnx quantization, and adapting code for compatibility with the PaddlePaddle framework. Furthermore, the user provided improvements to the test suite and examples within the repository.
berternieopenmmlabpaddleslimsparsity
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.