Q: Where did you go to school and when did you know you wanted to be a machine learning researcher?
A: I received my Masters degree from the Electrical and Computer Engineering department at UCSD, with a specialization in Intelligent Systems, Robotics, and Controls.
Coming to the United States as refugees with a family of 5 children, my parents had limited ability to send me to the best schooling institutions. Regardless, since a young age, I have always been adept at mathematical reasoning; my high school even needed to open a Calculus BC so that another classmate and I could progress in our mathematical curiosities. It was a “class” of 2.
After starting university I quickly realized an aptitude for Physics. Combining math and physics made engineering a good fit. I randomly choose Electrical Engineering and progressed in my education without much passion. In my Junior year I took a few courses focused on optimization and fell in love with Machine Learning. I applied to a masters program specializing in ML at UCSD. Given my new found passion and the opportunity to study the field, I devoted the next few years to becoming the best ML engineer / researcher I could. Since graduating, I’ve worked at Cerebras on some fascinating problems and I really enjoy what I do.
Q: What attracted you to Cerebras and what do you love the most about working here?
A: Having worked in ML for the past few years, what most intrigued me about Cerebras is the promise of the audacious goal it had put before itself. That goal: to accelerate A.I. 1000 fold.
Researchers can create very complex models. Unless hardware can actually run those models, what is the point? The promise of unlocking the ML community’s ability to expand experimentations at a massive scale was what attracted me. In 2012 Alex Krizhevsky (inventor of AlexNet) unlocked a 20x increase over what was possible for machine learning research. That innovation has created a resurgence in research and driven great innovation in technology.
I cannot wait to see what type of innovation we unlock with our technology. That audacious goal is why I came to Cerebras and what excites me most about working here.
Q: What are you working on today?
A: I work in the ML Team. I am currently looking at parallelization schemes in neural networks, neural network optimizers, as well as new neural architectures.
My last public facing project was published at NeurIPS 2019: Online Normalization for Training Neural Networks (https://arxiv.org/abs/1905.05894) . I will be presenting this work along with my colleagues and co-authors in Vancouver next week, and I couldn’t be more excited! – https://nips.cc/Conferences/2019/Schedule?showEvent=13906
Q: What has been your most rewarding and challenging project?
A: “Nothing in this world is worth having or worth doing unless it means effort, pain, difficulty” – Theodore Roosevelt
The most rewarding projects are also the most challenging. One of the things I like about working at Cerebras is that I am challenged by what I do. As a result, I get to be surprised by the person I get to become.
“O, do not pray for easy lives. Pray to be stronger men `and women`! Do not pray for tasks equal to your powers. Pray for powers equal to your tasks! Then the doing of your work shall be no miracle. But you shall be a miracle.” – Phillip Brooks
Q: Who are your most important teachers?
A: My parents have always been a moving force in my life. Besides teaching me to walk and talk, I think one of the greatest contribution in the making of Vitaliy has been their effort to instill in me a sense of hard work and perseverance. After immigrating out of the former Soviet Union, they taught me that I could achieve any version of the American dream as long as I was willing to work for it.
Apart from those amazing individuals, I’d like to think that my greatest teachers are all around me and it is my responsibility to learn from anyone and everyone who I come in contact with. I’m lucky to be at Cerebras, surrounded by a wealth of talented and driven individuals from whom I can absorb a world of knowledge.