Austin Huang
Senior Software Developer at Bloomberg
United States
Summary
🤩
Rockstar🎓
Top SchoolAustin Huang is a Senior Software Developer with nine years of experience building high-performance backend systems across finance and cloud platforms, currently at Bloomberg. His background includes multiple roles at Amazon and Broadway Technology, where he focused on production-grade software engineering and low-latency systems. Austin contributes to open source low-level ML tooling, including performance and instruction-tuning improvements to a lightweight C++ inference engine for Google's Gemma models. He combines strong C++ systems skills with practical DevOps experience from earlier roles, enabling reliable deployment and optimization of inference workloads. Based in the United States and trained at the University of Waterloo, he brings a pragmatic engineering approach that balances algorithmic efficiency with production constraints. Colleagues describe him as someone who surfaces subtle performance wins—like KV cache preallocation and tuning constants—that materially improve runtime behavior.
9 years of coding experience
4 years of employment as a software developer
Undergraduate, Computer Science, Undergraduate, Computer Science at University of Waterloo
English, French, Chinese
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (8)
tensorrt10
machine-learning10
c-language10
tensor10
tensorflow10
cprogramming-language10
operation10
performance-optimization9
Programming languages (12)
Github contributions (5)
google/gemma.cpp
Feb 2024 - Sep 2024
lightweight, standalone C++ inference engine for Google's Gemma models.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:2 releases, 30 reviews, 16 PRs in 7 months
Contributions summary:Austin primarily worked on improving the C++ inference engine for Google's Gemma models. Their commits focused on modifying core files like `ops.h`, `run.cc`, and `gemma.h` to refine and optimize the inference process. These changes include updates to the preallocation size of the KV cache, as well as improvements to the code formatting and the addition of performance-related constants. The contributions also added functionality for instruction-tuned models.
hasktorch/ffi
Sep 2017 - Dec 2018
The hasktorch FFI bindings to the ATen library
Contributions:94 commits, 5 pushes in 1 year 3 months
rustatenffiffi-bindings
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.