Tianqi Chen
Distinguished Engineer at NVIDIA
Seattle, Washington, United States
Summary
🤩
Rockstar🎓
Top School
Top expert inHigh-Performance Machine Learning Computing
Tianqi Chen is a Seattle-based Distinguished Engineer at NVIDIA and an Assistant Professor at Carnegie Mellon University, combining academic rigor with industry leadership across 11 years in machine learning systems. A co-founder of OctoML, he is a prolific open-source maintainer and architect behind foundational projects such as XGBoost (rabit/MPI), DLPack (the universal tensor ABI), TVM and MXNet, focusing on scalable distributed training, compiler-level optimizations, and cross-platform tensor interoperability. His contributions range from low-level IO, memory allocators and MPI integration to IR/TIR refactors and deployment tooling, showing a rare fluency in both performance-critical backend engineering and production deployment. Trained at the University of Washington, he pairs deep systems expertise with hands-on work across back-end C++ stacks and front-end LLM deployments, often driving subtle but high-impact infrastructure changes.
12 years of coding experience
5 years of employment as a software developer
Doctor of Philosophy (PhD) Computer Science, Doctor of Philosophy (PhD) Computer Science at University of Washington
Github Skills (82)
graph-algorithms10
operation10
c1110
c1710
lua10
deep-learning10
parallel-computing10
webgpu10
c-plus-plus10
javascript10
convolutional-neural-networks10
machinelearningmodel10
memory-management10
machinelearning10
neural-network10
Programming languages (20)
Github contributions (5)
mlc-ai/mlc-llm
Apr 2023 - Mar 2025
Universal LLM Deployment Engine with ML Compilation
Role in this project:
Back-end Developer
Back-end DeveloperContributions:2 releases, 381 reviews, 745 PRs in 1 year 10 months
Contributions summary:Tianqi's primary contribution is the addition of a command-line interface (CLI) feature to the project, specifically a "stats" command. This feature allows users to view statistics from the latest round of conversation within the application. The implementation included modifications to the `cpp/cli_main.cc` file to incorporate the new command and its functionality, utilizing functions from the project’s chat module. The user also added support for loading and using external tokenizer libraries.
language-modelllmmachine-learning-compilationtvm
dmlc/dmlc-core
Mar 2015 - Oct 2020
A common bricks library for building scalable and portable distributed machine learning.
Role in this project:Back-end Developer
Back-end DeveloperContributions:1 release, 4 reviews, 403 commits in 5 years 7 months
Contributions summary:Tianqi primarily worked on the I/O components of the `dmlc-core` library. Their contributions involved implementing interfaces for stream I/O, including `IStream`, `ISeekStream`, and `InputSplit`. The user created specific implementations like `FileStream`, `HDFSStream`, and `LineSplitter`, and refactored existing code to accommodate changes, added endian aware serialization, and allowed support for text format data for the IO system. The user also worked on refactoring of the parser with more performance
scalablebricksmachine-learningportabledistributed-machine-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.