I’m an engineer by training, working on training.
My technical interests are across the stack: machine learning frameworks, distributed systems, and all things GPUs. I enjoy efficiency problems, understanding what makes a system slow and helping people get more out of less.
Currently, I work at Applied, where my team and I build ML infrastructure for training and serving autonomy models. Most recently, I led our GPU efficiency efforts, accelerating throughput for our end-to-end model training architecture by 10x.
Before that, I helped build and scale Cloud Engine, a platform to help companies continuously validate their autonomous vehicles in simulation. There, I reduced query latency in our results search platform by >99%, enabling users to analyze millions of simulation results in milliseconds instead of minutes. This effort later scaled organization-wide to triage and optimize costly queries.
I’ve also worked at Productiv, Amazon, and contributed to Ray.
I graduated from UC Berkeley with a B.A. in Computer Science and a B.A in Data Science. At Cal, I spent much of my time developing software for non-profits, teaching databases, and conducting data systems research.
Outside of work, I enjoy home-cooking, reading, and fighting people. View my resumé or shoot me an email at micahtyong@gmail.com.