Leapfrogging GPU Servers: Systems for Cost-Effective Scaling of LLM Inference, from First Principles | Kisaco Research

Leapfrogging GPU Servers: Systems for Cost-Effective Scaling of LLM Inference, from First Principles

Session Topics: 
Generative AI
Systems
Infrastructure
Sponsor(s): 
Positron AI
Speaker(s): 

Author:

Thomas Sohmers

Founder and CEO
Positron AI

Thomas Sohmers is an innovative technologist and entrepreneur, renowned for his pioneering work in the field of advanced computing and artificial intelligence. Thomas began programming at a very early age, which led him to MIT as a high school student where he worked on cutting-edge research. By the age of 18, he had become a Thiel Fellow, marking the beginning of his remarkable journey in technology and innovation. In 2013, Thomas founded Rex Computing, where he designed energy-efficient processors for high-performance computing applications. His groundbreaking work earned him numerous accolades, including a feature in Forbes' 30 Under 30. After a stint exploring the AI industry, working on scaling out GPU clouds and large language models, Thomas founded and became CEO of Positron in 2023. Positron develops highly efficient transformer inferencing systems, and under Thomas's leadership, it has quickly become one of the most creative and promising startups in the AI industry.

... read more

Thomas Sohmers

Founder and CEO
Positron AI

Thomas Sohmers is an innovative technologist and entrepreneur, renowned for his pioneering work in the field of advanced computing and artificial intelligence. Thomas began programming at a very early age, which led him to MIT as a high school student where he worked on cutting-edge research. By the age of 18, he had become a Thiel Fellow, marking the beginning of his remarkable journey in technology and innovation. In 2013, Thomas founded Rex Computing, where he designed energy-efficient processors for high-performance computing applications. His groundbreaking work earned him numerous accolades, including a feature in Forbes' 30 Under 30. After a stint exploring the AI industry, working on scaling out GPU clouds and large language models, Thomas founded and became CEO of Positron in 2023. Positron develops highly efficient transformer inferencing systems, and under Thomas's leadership, it has quickly become one of the most creative and promising startups in the AI industry.