Aaron Wang
Hello, my name is Aaron (legal name Junyao), and I’m a computer enginnering student at the University of Waterloo. I enjoy working on and learning about system-level software, performance engineering, and robotics.
Previously, I interned at CentML as an Machine Learning Systems Engineer, where I helped optimize LLM inferencing. I worked on features such as dynamic speclative decoding, gRPC runtime environment, and model compilation. Also, I was a research intern at Huawei on the AI infrastructure team. I helped research optimization within AI systems, especially on the distributed training side. I co-authored a paper (currently under review!) on collective communication scheduling algorithms in GPU clusters.
On the side, I also help lead & write drone software at the Waterloo Aerial Robotics Group. We build anything software-related to help drones fly, from computer vision models to full-stack ground station software. In my free time, I enjoy climbing, skiing, sampling local ramen restaurants, and playing video games.
In 2025, I plan to document all my past projects in the blog section!
Past Experiences
September 2024 - December 2024
Working on CentML's LLM inference engine! Adding performance optimizations and improving tooling.
January 2024 - April 2024
Researching ways to improve large AI training clusters on the system level, co-authored a paper on collective-communication optimizations
May 2023 - August 2023
Prototyped webpages and dashboards, learned a lot about Full-Stack development