Kunlun Li
Software Engineer at NVIDIA
Shanghai, Shanghai, China
Summary
🤩
RockstarKunlun Li is a software engineer based in Shanghai with four years of experience focused on high-performance GPU-accelerated deep learning systems. Currently at NVIDIA, he contributes to the HugeCTR project, improving model parallelism and integrating sparse embedding solutions like the Sparse Operation Kit for recommendation model benchmarks. He brings practical expertise in distributed training with TensorFlow, container configuration, and compilation fixes that bridge research prototypes to production-ready frameworks. Notably, his work targets CTR estimation workloads, a niche requiring careful optimization of sparse operations and GPU resources. Colleagues benefit from his hands-on approach to debugging complex build and deployment issues in large open-source ML codebases.
4 years of coding experience
Github Skills (8)
recommendation-system10
deep-learning10
cpp10
tensorflow10
gpu-acceleration10
distributed-training9
cuda8
mlops7
Programming languages (3)
Github contributions (5)
NVIDIA-Merlin/HugeCTR
Sep 2021 - Oct 2022
HugeCTR is a high efficiency GPU framework designed for Click-Through-Rate (CTR) estimating training
Role in this project:
ML Engineer
ML EngineerContributions:1 review, 20 commits, 4 pushes in 1 year
Contributions summary:Kunlun's contributions primarily focus on enhancing the Hugectr framework for high-efficiency GPU-accelerated deep learning, especially in the context of Click-Through-Rate (CTR) prediction. They implemented a model parallelism demo using TensorFlow's native methods, demonstrating expertise in distributed training techniques. Their work involves modifying and integrating components related to sparse embedding operations, along with adapting the framework to utilize the SOK (Sparse Operation Kit) for DLRM (Deep Learning Recommendation Model) benchmarks, including fixing compilation issues and updating container configurations.
cudapytorchcppgpu-accelerationdeep-learning
kunlunl/TransformerEngine
Dec 2023 - Mar 2025
Contributions:67 pushes, 9 branches in 1 year 3 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.