Sasha Sobol
Member Of Technical Staff at Plato
Sunnyvale, California, United States
Summary
🤩
Rockstar🎓
Top SchoolSasha Sobol is a Member of Technical Staff based in Sunnyvale with 11 years building distributed systems and infrastructure for AI and autonomous-vehicle startups and tech giants. His career spans Google, drive.ai, Apple, Anyscale, and now Plato, where he blends backend engineering with DevOps to make large-scale ML and runtime environments reliable and reproducible. Notably, he contributed to the widely used Ray project—improving the autoscaler, placement-group resource handling, and SGD v2 prototype—demonstrating deep expertise in cluster management and scalable workload orchestration. Sasha moves comfortably between product-focused engineering and low-level infrastructure work, shipping integration tests and documentation as rigorously as features. Colleagues rely on him for pragmatic solutions that tame complex resource allocation and node lifecycle challenges in production.
11 years of coding experience
18 years of employment as a software developer
57
Github Skills (14)
autoscaling10
ray10
distributed-systems10
python10
yaml9
deploying8
testing8
githubaction-workflow7
dockers7
kubernetes-pods7
kubernetes7
docker7
github-ci7
machine-learning6
Programming languages (6)
Github contributions (5)
ray-project/ray
Aug 2021 - Oct 2021
Ray is an AI compute engine. Ray consists of a core distributed runtime and a set of AI Libraries for accelerating ML workloads.
Role in this project:
Back-end Developer & DevOps Engineer
Back-end Developer & DevOps EngineerContributions:75 reviews, 8 commits, 13 PRs in 2 months
Contributions summary:Sasha contributed primarily to the Ray autoscaler component, implementing features such as enforcing per-node-type max workers. They also addressed resource allocation within placement groups and added annotations for API stability. The user's work involved supporting streaming output for runtime environment setup and contributing to the development of the SGD v2 prototype. Furthermore, the user worked on configurations related to node management, including setting default min/max workers, updating documentation, and adding integration tests.
pythonconsistsruntimetensorflowserving
sasha-s/go-deadlock
Jul 2016 - Sep 2021
Contributions:3 releases, 5 reviews, 35 commits in 5 years 2 months
golangdeadlockdeadlock-detectionmutexonline-deadlock-detection
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.