Tristan Heywood
Member Of Technical Staff at OpenAI
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolTristan Heywood is an AI-focused software engineer and researcher with seven years of experience, currently a Member of Technical Staff at OpenAI working on accelerating LLM inference via custom Triton kernels. He graduated top of his cohort with University Medals in Mechatronic Engineering and Computer Science and has a track record of shipping high-performance robotics and perception code—most notably a C++ topological navigation system developed during an honours stint at NASA JPL for the DARPA Subterranean Challenge. Tristan has blended research and product experience across startups and labs, tuning object-detection models, building training/inference pipelines, and contributing ML components to open-source point-cloud frameworks like torch-points3d. Comfortable across C++, Python, PyTorch and full-stack tooling, he brings both low-level performance optimization skills and practical deployment experience from ag‑tech to large-scale AI systems. An unexpected thread through his work is a knack for translating geometric deep learning research into robust production components, evidenced by contributions to RandLA-Net and ResNet-style blocks in PyTorch.
7 years of coding experience
2 years of employment as a software developer
The University of Sydney
Github Skills (8)
mle10
pytorch10
deep-learning10
point-cloud10
ml10
deeplearning-ai9
image-segmentation8
segmentation8
Programming languages (5)
Github contributions (5)
torch-points3d/torch-points3d
Dec 2019 - Feb 2020
Pytorch framework for doing deep learning on point clouds.
Role in this project:
ML Engineer
ML EngineerContributions:59 commits, 14 PRs, 145 pushes in 2 months
Contributions summary:Tristan primarily focused on implementing and integrating deep learning components within the PyTorch framework. Their work involved defining and implementing a ResNet block base class, moving sampling functions within core modules, and developing a dilated residual block for the RandLANet model. They also made minor adjustments to the training script, contributing to the overall model architecture and training process.
pytorchscannetpytorch-frameworkpoint-cloudss3dis
tristanheywood/colorimetric_sensor_reader
Jan 2021 - Jul 2021
Contributions:2 releases, 2 PRs, 37 pushes in 6 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.