Travis Addair
Senior Director Of Engineering, AI at Rubrik
Seattle, Washington, United States
Summary
🤩
Rockstar🎓
Top SchoolTravis Addair is a senior engineering leader and AI systems builder with nine years of focused experience delivering distributed deep learning infrastructure and production model-serving platforms. As Co-Founder and CTO of Predibase and now Senior Director of Engineering, AI at Rubrik, he has driven innovations that make customized LLMs practical and cost-effective—such as LoRAX and Turbo LoRA adapters that enable serving hundreds of task-specific models on a single GPU. He is a long-time maintainer of Horovod and co-maintainer of Ludwig, contributing core fixes that make distributed training differentiable and robust across TensorFlow and PyTorch ecosystems. His work spans low-level optimizer integrations, multi-GPU synchronization, and Kubernetes-backed ML infrastructure, reflecting both research-grade ML expertise and production-grade DevOps skill. Based in Seattle and active on GitHub, he combines hands-on engineering with product leadership to move cutting-edge model training and serving from prototype to scale. An unexpected throughline in his career is a background in game AI and large-scale scientific data pipelines, which informs his pragmatic approach to performant, resilient ML systems.
9 years of coding experience
19 years of employment as a software developer
Bachelor of Science (B.S.) Computer Science Mathematics (minor), Bachelor of Science (B.S.) Computer Science Mathematics (minor) at Northern Arizona UniversityMaster of Science (M.S.) Computer Science Artificial Intelligence, Master of Science (M.S.) Computer Science Artificial Intelligence at Stanford University
Stackoverflow
427reputation
37kreached
13answers
2questions
Github Skills (31)
kubernetes10
pytorch10
distributed-training10
python10
data-science10
machine-learning10
pytorch-lightning10
horovod10
deep-learning10
tensorflow10
kubernetes-pods10
ray10
devops10
gradient-descent9
unit-testing9
Programming languages (9)
Github contributions (5)
horovod/horovod
Jul 2018 - Oct 2022
Distributed training framework for TensorFlow, Keras, PyTorch, and Apache MXNet.
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:14 releases, 419 reviews, 565 commits in 4 years 4 months
Contributions summary:Travis's contributions primarily revolved around enhancing the TensorFlow API for Horovod. They made Horovod operations differentiable in TensorFlow, implemented unit tests to verify gradient correctness, and fixed several issues related to gradients for allgather and broadcast operations. Furthermore, the user addressed compatibility issues with specific TensorFlow versions, demonstrating a focus on both framework integration and the robustness of Horovod's core functionality.
distributed-trainingkerasmxnetpytorchtensorflow
ludwig-ai/ludwig
Sep 2019 - Jan 2023
Low-code framework for building custom LLMs, neural networks, and other AI models
Role in this project:Back-end Developer
Back-end DeveloperContributions:5 releases, 1186 reviews, 675 commits in 3 years 5 months
Contributions summary:Travis's commits primarily involve the modification and optimization of optimizers within the Ludwig framework. This includes creating an optimizer wrapper, adapting the framework to support TensorFlow optimizers, and addressing Horovod integration. The contributions reflect an effort to improve training performance and distributed training capabilities, specifically supporting the incorporation of external components.
ailow-codeneural-networkdeep-learningmachine-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.