Leslie Fang
Software Developer at Intel
Chandler, Arizona, United States
Summary
🤩
RockstarLeslie Fang is a software developer and PyTorch engineer at Intel with six years of experience optimizing deep learning workloads for Intel architectures. She focuses on backend and ML engineering, contributing performance-critical enhancements to projects like intel-extension-for-pytorch and core PyTorch, including fused quantized operators and dynamic-shape improvements. At Intel she has driven BF16, FP32 and INT8 inference support and refactored reference models (e.g., SSD-ResNet34) to improve throughput and accuracy on Xeon and Data Center GPUs. Her work spans both library-level operator optimizations (MKL-DNN integration, In- ductor C++ wrapper) and practical tutorials that help others apply quantization and mixed-precision techniques. Colleagues rely on her to translate low-level numerical and dtype issues into robust, high-performance implementations. Based in Chandler, Arizona, she combines production-grade engineering with open-source stewardship in widely used ML infrastructure.
6 years of coding experience
Github Skills (22)
pytorch10
c-language10
machine-learning10
deep-q-learning10
inference10
vectorization10
deep-learning10
tensorflow10
performance-optimization10
quantization10
cprogramming-language10
intel9
cpu9
documentation9
cuda9
Programming languages (5)
Github contributions (5)
pytorch/pytorch
Apr 2021 - Jan 2023
Tensors and Dynamic neural networks in Python with strong GPU acceleration
Role in this project:
ML Engineer
ML EngineerContributions:1122 reviews, 375 commits, 375 PRs in 1 year 9 months
Contributions summary:Leslie's contributions primarily focused on enhancing the PyTorch framework, specifically within the realm of quantized neural networks. Their work involved fixing data type issues related to Conv-BN folding and handling of different data types in quantized layers. They added and optimized several fused quantized operators like `conv2d_add`, `conv2d_add_relu`, etc., which can improve the inference performance. Furthermore, their contributions included work on dynamic shape support and various pattern matching improvements to the graph compilation for quantized models.
pythongpu-accelerationdeep-learninggpunumpy
intel/intel-extension-for-pytorch
Mar 2021 - Sep 2022
A Python package for extending the official PyTorch that can easily obtain performance on Intel platform
Role in this project:Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:65 commits, 1 comment in 1 year 6 months
Contributions summary:Leslie primarily contributed to the development and enhancement of the `intel-extension-for-pytorch` library. The contributions involve the implementation and optimization of PyTorch operations, specifically focusing on performance improvements for Intel platforms. Key contributions include the addition of support for code free optimization, and the addition of custom operations related to inference and the integration of MKL-DNN for accelerated performance.
pytorchpythondeep-learningintelmachine-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.