Kaokao Lv
Machine Learning Engineer at Intel Corporation
Shanghai, Shanghai, China
Summary
🤩
RockstarKaokao Lv is a Machine Learning Engineer based in Shanghai with seven years of experience building and optimizing ML systems for production. Currently at Intel, Kaokao focuses on model compression and practical deployment, contributing to high-impact open-source work such as Intel's neural-compressor where they fixed quantization tuning issues, added unit tests, and implemented a diffusion model example. Skilled in deep learning, data mining, and large-scale systems like Hadoop and Spark, they bridge research and engineering to make state-of-the-art techniques (INT8/FP8/INT4/NF4 quantization and sparsity) more usable across TensorFlow, PyTorch, and ONNX Runtime. Known for attention to detail in both code and documentation, Kaokao brings a pragmatic, production-first mindset to pushing ML models toward efficiency and reliability.
7 years of coding experience
Github Skills (8)
quantization10
pytorch10
mxnet10
large-language-models9
auto-tuning9
machine-learning8
python8
documentation7
Programming languages (3)
Github contributions (5)
intel/neural-compressor
Sep 2022 - Dec 2022
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:28 reviews, 13 commits, 7 PRs in 3 months
Contributions summary:Kaokao contributed to the `intel/neural-compressor` repository by addressing issues and adding new features. Their work includes fixing a tuning history issue and adding a unit test related to quantization strategy. Furthermore, they addressed a bug related to calling Tensor objects and implemented a diffusion model example. Additionally, the user worked on documentations for MXNet dataloader and util functions.
knowledge-distillationauto-tuningcompressorsparsityintel
lkk12014402/optimum-habana
Jan 2024 - Feb 2025
Easy and lightning fast training of 🤗 Transformers on Habana Gaudi processor (HPU)
Contributions:60 pushes, 9 branches in 1 year
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.