Xiaodong Huang
Deep Learning Software Engineer at NVIDIA
California, United States
Summary
🤩
Rockstar🎓
Top SchoolXiaodong Huang is a Deep Learning Software Engineer with 11 years of experience designing and optimizing GPU-driven ML and graphics systems at NVIDIA, currently focused on TensorRT for high-performance deep learning inference. His career progressed from graphics and GPU verification tools to CUDA kernel analysis and finally to architecting inference optimizations and engine builds for transformer models. He has hands-on experience improving TensorRT performance—disabling tactics, integrating timing caches, and fixing quantization and masking issues—to ensure robust, cross-architecture inference. Based in California, Xiaodong pairs deep low-level GPU expertise with practical ML deployment work, bridging driver-level insights and production inference engineering. A Master of Engineering from Fudan University, he brings a rare combination of compiler/tooling, kernel-level performance tuning, and model-focused optimization that accelerates real-world AI workloads.
11 years of coding experience
15 years of employment as a software developer
Master of Engineering, Master of Engineering at Fudan UniversityBachelor, Bachelor at Southeast University
Github Skills (8)
tensorrt10
inference10
deep-learning10
gpu-acceleration10
bert10
pytorch8
nvidia8
python8
Programming languages (6)
Github contributions (5)
NVIDIA/TensorRT
Jul 2020 - Oct 2022
NVIDIA® TensorRT™ is an SDK for high-performance deep learning inference on NVIDIA GPUs. This repository contains the open source components of TensorRT.
Role in this project:
ML Engineer
ML EngineerContributions:1 release, 40 reviews, 31 commits in 2 years 4 months
Contributions summary:Xiaodong primarily contributed to optimizing and improving the performance of deep learning inference within the NVIDIA TensorRT framework. Their work included enhancing the engine build process for BERT models, specifically by disabling certain tactics and integrating timing caches. They also addressed bugs and made code adjustments related to the use of full masks for specific hardware and data types, ensuring optimal performance across different architectures. In addition, they provided updates and fixes related to pytorch-quantization.
deep-learninginferencenvidiatensorrtgpu-acceleration
ttyio/flashinfer
Jul 2025 - Sep 2025
FlashInfer: Kernel Library for LLM Serving
Contributions:58 pushes, 20 branches in 2 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.