Greg Diamos


Greg's mission is to build complex systems that the world can use.

He had previously co-founded MLPerf & MLCommons, which have now set the industry standards for high performance machine learning systems.

He had led LLM scaling at Baidu, where his team published the first paper on scaling laws. He deployed one of the first LLMs into production to 100M people in 2009 in the Baidu search engine, now used by billions.

Greg was also an architect — or CUDA Ninja — at NVIDIA in the very early days of deep learning (2008), where he contributed key inventions to the Volta GPU and worked on the first AI project there.

He holds a PhD from the Georgia Institute of Technology in high-performance computing.