Suggestions
Anton Polishko, PhD
SWE endgame: LLM doing Rust macros
Anton Polishko, PhD is a technology professional and entrepreneur based in the San Francisco Bay Area. He is currently the Founder of OpenBabylon, a company he established recently.12
Professional Experience
Anton has a diverse background in software engineering, artificial intelligence, and academic research. Some key highlights of his career include:
- Software Engineer at Google from September 2021 to March 20231
- Director of Cloud Infrastructure and AI researcher at CoronaWhy from March 2020 to September 20211
- Software Engineer at Helpful Bus from September 2014 to March 20201
- CTO at ZULLOO™ from October 2015 to February 20181
Academic Background
Anton Polishko holds a strong academic foundation:
- PhD from the University of California, Riverside (2009-2014)1
- Master of Science with Honors from Kyiv National Taras Shevchenko University (2007-2009)1
- Bachelor of Science with Honors from Kyiv National Taras Shevchenko University (2004-2007)1
During his time at UC Riverside, he also served as a Graduate Student Researcher and Graduate Teaching Assistant.1
Entrepreneurial Endeavors
Besides his current role as Founder of OpenBabylon, Anton has shown an interest in entrepreneurship and startup ecosystems:
- Participated in Startup School Online in 20171
- Attended the Unbank.Ventures FinTech Incubator in 20171
- Served as a Mentor at Founder Institute from May 2016 to March 20201
Anton Polishko's diverse experience in technology, research, and entrepreneurship has likely contributed to his current role as the Founder of OpenBabylon, a relatively new venture in his career.
Highlights
The future is here, old man!
It's ridiculous that it's possible in less than few hours (time it takes to download 231Gb of q4_0 weights) to launch 405B parameter LLM beating GPT-4 running on a single node. https://lnkd.in/gBW5Trae
I still remember running llama:65B at twice slower t/s eval rate just last year.
I’m proud to be on the same list of authors with folks from Google Brain and YouTube Music DevX teams.
It was truly peak experience at Google and required lots of deep work time from me to 3x speed up agent simulations on Youtube infrastructure and scale.
