Applied research, post-training, reinforcement learning
I am a Senior Director of Applied Research at NVIDIA where I lead post-training teams responsible for Nemotron post-training. Before that I helped grew NVIDIA's NeMo from 1 person (myself) into a multi-team effort.
Before NVIDIA, I worked as applied scientist at Apple, working on AppStore App and Games recommendations. And before that I worked at Microsoft where I wrote first version of GPU backend for CNTK.
I received my Ph.D. (in 2010) an and M.Sc. (in 2009) degrees in Information and Computer Science from University of California, Irvine. My advisor was: Prof. Natasa Przulj. I also go M.Sc. (in 2007) and B.Sc. (in 2005) in Applied Mathematics from National Taras Shevchenko University of Kyiv, Ukraine.
The best way to reach me is via kuchaev at gmail dot com.
Notes on applied research, systems, and modeling.
Recently, I gave a short talk about NeMo at MLSys workshop at NeurIPS 2019 in Vancouver, Canada.NeMo is the project I actively work on at NVIDIA. It is open-source and we welcome external contributors. Get it from github. Slides Live link
Read post