After I graduated from South China University of Technology, I joined Intel and worked there for a year as an AI framework engineer. I was a part of the graph compiler team. Our product is an acceleration library of oneDNN, which provides transparent acceleration with Intel’s server chips. While I was there, it was under pre-release, but now it has been fully integrated into oneDNN. You can find the release of the product on the oneDNN GitHub repository.

At Intel, I undertook a variety of roles. Initially, I was involved in integrating bf16 into our graph compiler. Later, I worked on implementing some graph operators in our compiler. I then proceeded to implement the computation graph for MHA and monitor its performance.

Throughout the year, my manager, Zhennan Qin, and my mentor, Yijie Mei, were incredibly supportive. I gained invaluable insights into the nuances of software engineering in the industry, learning about industry standards and the processes involved in software development. Familiarizing myself with Continuous Integration (CI), collaborating on GitHub, and understanding code linting and the make system were among the key takeaways. This real-world experience differed significantly from what I was taught during my undergraduate years.

I resigned in August 2022 to pursue further studies at Boston University.