Sign In

Vikram Sreekanti

Co-founder & CEO, RunLLM

Vikram Sreekanti is an accomplished professional with a solid educational background and diverse experience in the field of computer science and software engineering.

He completed his Bachelor of Science in Electrical Engineering and Computer Science followed by a Doctor of Philosophy in Computer Science at the prestigious University of California, Berkeley.

Vikram has been actively involved in various organizations where he has made significant contributions. He served as a Graduate Researcher at RISE Lab, UC Berkeley, showcasing his expertise and commitment to research and development.

Prior to his role at RISE Lab, he gained practical experience as a Software Engineer at AMPLab, UC Berkeley, honing his skills in software development and engineering.

Additionally, Vikram has also interned at renowned companies such as Cloudera and Yammer, Inc., further enhancing his knowledge and capabilities in the tech industry.

His professional journey reflects a continuous dedication to learning, research, and practical application of computer science principles, making him a valuable asset in the field.

Highlights

May 27 · twitter

With all the focus being on model releases for the last couple years, the last mile of AI applications has received relatively little attention.

"Can we find the right answer?" is hard enough, but “can we find the right answer, in time, for the person who needs it?” is even more important.

More in the post below 👇

May 26 · twitter

A year ago, I was skeptical about open-source LLMs — I thought they’d stay behind the frontier forever. I was wrong.

I was pretty skeptical that open-source models weren't going to be able to compete at the frontier of model innovation. We even wrote about this a few times on the blog.

There were two reasons why we were wrong.

First, the process of pre-training itself has gotten more efficient which means that teams are able to do more with less.

Second, more and more effort is being put into post-training, which seems to be a more scalable approach, especially as we get better and better fine-tunes of open models.

Generally, it feels like the rate of innovation at the frontier has slowed, which makes it easier to catch up.

Vikram Sreekanti
Add to my network

Location

San Francisco Bay Area