Andrew Aikawa
Co-Founder, CTO at Trainy
Berkeley, California, United States
Summary
👤
Senior🎓
Top SchoolAndrew Aikawa is a CTO and co-founder focused on modernizing GPU infrastructure and squeezing more ROI out of AI training workloads, currently leading engineering at Trainy (YC S23). With a PhD in Physics from UC Berkeley and nine years of hands-on experience, he blends deep research pedigree with practical MLOps and distributed training expertise. He’s shipped production-scale training and serving pipelines—running large distributed training on hundreds of A100s, optimizing custom CUDA/Torch ops, and deploying models via Triton for real-time inference. An active open-source contributor, Andrew has improved SkyPilot’s multi-cloud and Kubernetes GPU integrations, enabling easier, cost-efficient AI job execution across clouds. He’s particularly interested in tooling that boosts multinode throughput, and brings an uncommon mix of low-level GPU systems work and applied ML production experience.
9 years of coding experience
1 year of employment as a software developer
Doctor of Philosophy - PhD, Physics, Doctor of Philosophy - PhD, Physics at University of California, Berkeley
Github Skills (14)
digitalocean10
kubernetes10
infrastructure10
cloud-infrastructure10
cloud-computing10
python10
kubernetes-pods10
dockers9
automation-testing9
automation9
automations9
docker9
mlops8
gpu8
Programming languages (7)
Github contributions (5)
skypilot-org/skypilot
Jul 2023 - Dec 2024
The AI Compute Platform for frontier teams. SkyPilot turns fragmented AI compute into one AI supercomputer, so frontier AI teams build custom intelligence faster.
Role in this project:
Cloud Engineer / Infrastructure Engineer
Cloud Engineer / Infrastructure EngineerContributions:43 reviews, 25 PRs, 95 comments in 1 year 6 months
Contributions summary:Andrew's commits primarily focus on integrating and supporting the SkyPilot project with different cloud platforms, specifically Paperspace and Kubernetes. They were actively involved in adding and improving Paperspace cloud integration by implementing features such as launch, stop, and autodown functionality. Moreover, the user contributed to improvements related to GPU feature discovery in Kubernetes, and updated core parameters for Kubernetes, and generally improved the integration of SkyPilot with Kubernetes. Additionally, they addressed issues such as improving timeouts and correcting SSH setups.
aisupercomputercloud-computingdeep-learninggpu
Trainy-ai/konduktor
May 2024 - Feb 2025
cluster/scheduler health monitoring for GPU jobs on k8s
Contributions:15 reviews, 123 PRs, 141 pushes in 9 months
gpuk8smonitoringaideep-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.