|
Evan Racah
ejracah at gmail dot com
I began my career at NERSC at Berkeley Lab, applying deep learning to climate and high-energy physics while scaling machine learning algorithms on supercomputers. I then earned my master’s at Mila, working with Chris Pal and Yoshua Bengio on unsupervised visual representation learning.
Next, I joined the Waymo R&D team to integrate deep learning and RL into autonomous vehicles. Afterward, as a research software engineer at MosaicML(acquired by Databricks), I helped build the LLM pretraining library, Composer, focusing on distributed training and checkpointing capabilities that later powered the training of DBRX.
Google Scholar / Github / Twitter / CV / Blog
|
|
|
Research
I am very interested in embracing my AI for science roots and getting back into deep learning for science, specifically biological applications. I am fascinated by designing new proteins and molecules using structural ML approaches as well as harnessing multi-modal self-supervised self-supervised approaches on sequential, structural, and visual data to learn new representations that can facilitate treating diseases such as cancer.
|
|
